Fall risk assessment method, fall risk assessment device, and recording medium

文档序号：928503 发布日期：2021-03-05 浏览：2次中文

阅读说明：本技术 跌倒风险评估方法、跌倒风险评估装置以及记录介质 (Fall risk assessment method, fall risk assessment device, and recording medium ) 是由樋山贵洋佐藤佳州相原贵拓和田健吾滨塚太一松村吉浩于 2020-08-26 设计创作，主要内容包括：本发明提供一种基于试验对象的步行动作评估跌倒风险的跌倒风险评估装置的跌倒风险评估方法、跌倒风险评估装置以及记录介质。跌倒风险评估方法中,获取与试验对象的步行相关的步行数据；根据步行数据,检测试验对象的腰部在试验对象的一条腿的立脚期在垂直方向的位移、试验对象的腰部在一条腿的脚悬空期在垂直方向的位移、一条腿的膝关节在立脚期的角度以及一条腿的脚踝关节在脚悬空期的角度的至少其中之一；利用腰部在立脚期在垂直方向的位移、腰部在脚悬空期在垂直方向的位移、膝关节在立脚期的角度以及脚踝关节在脚悬空期的角度的至少其中之一,判断试验对象的跌倒风险。(The invention provides a fall risk assessment method for a fall risk assessment device for assessing fall risk based on walking motion of a test subject, a fall risk assessment device, and a recording medium. In a fall risk assessment method, walk data relating to the walking of a test subject is acquired; detecting at least one of a vertical displacement of the waist of the test subject in the stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject in the foot suspension phase of one leg, an angle of the knee joint of one leg in the stance phase, and an angle of the ankle joint of one leg in the foot suspension phase, based on the walking data; and judging the falling risk of the test object by utilizing at least one of the displacement of the waist in the vertical direction in the foot standing period, the displacement of the waist in the vertical direction in the foot suspension period, the angle of the knee joint in the foot standing period and the angle of the ankle joint in the foot suspension period.)

1. A fall risk assessment method is a fall risk assessment method of a fall risk assessment apparatus for assessing fall risk based on walking motion of a test subject, and is characterized by comprising the following steps:

acquiring walking data related to the walking of the test subject;

detecting at least one of a vertical displacement of the waist of the test subject during the stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject during the foot suspension phase of the one leg, an angle of the knee joint of the one leg during the stance phase, and an angle of the ankle joint of the one leg during the foot suspension phase, based on the walking data;

and determining the falling risk of the test object by using at least one of the displacement of the waist in the vertical direction in the standing period, the displacement of the waist in the vertical direction in the foot suspension period, the angle of the knee joint in the standing period and the angle of the ankle joint in the foot suspension period.

2. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the displacement of the waist in a vertical direction during a predetermined period of the stance phase is detected,

in the determination, the fall risk of the test subject is determined using an average of the time-series data of the displacement of the waist in the vertical direction.

3. Fall risk assessment method according to claim 2,

when the period from the time when one foot of the test subject lands on the ground to the time when the foot lands on the ground again is expressed as one walking cycle and the one walking cycle is expressed by 1% to 100%,

the predetermined period is a period of 1% to 60% of the one walking cycle.

4. Fall risk assessment method according to claim 2,

the predetermined period is a period of 9% to 19% of the one walking cycle.

5. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the displacement of the waist in a vertical direction during a predetermined period of the foot suspension period is detected,

in the determination, the fall risk of the test subject is determined using an average of the time-series data of the displacement of the waist in the vertical direction.

6. A fall risk assessment method according to claim 5,

the predetermined period is a period of 61% to 100% of the one walking cycle.

7. Fall risk assessment method according to claim 1,

in the detecting, time-series data of an angle of the knee joint in a predetermined period of the stance phase is detected,

in the determining, the fall risk of the test subject is determined using an average value of the time-series data of the angle of the knee joint.

8. Fall risk assessment method according to claim 7,

the predetermined period is a period of 1% to 60% of the one walking cycle.

9. Fall risk assessment method according to claim 7,

in the determining, the fall risk of the test subject is determined using the angle of the knee joint at a time of 35% of the one walking cycle.

10. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the angle of the ankle joint during a predetermined period of the foot suspension period is detected,

in the determining, the fall risk of the test subject is determined using an average value of the time-series data of the angle of the ankle joint.

11. Fall risk assessment method according to claim 10,

the predetermined period is a period of 61% to 100% of the one walking cycle.

12. Fall risk assessment method according to claim 10,

the predetermined period is a period of 84% to 89% of the one walking cycle.

13. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the displacement of the waist in a vertical direction during a first period of the stance phase and time-series data of the angle of the knee joint during a second period of the stance phase are detected,

in the determining, the fall risk of the test subject is determined using an average value of time-series data of the displacement of the waist in the vertical direction during the first period and an average value of the time-series data of the angle of the knee joint during the second period.

14. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the displacement of the waist in the vertical direction during a first period of the foothold period, time-series data of the displacement of the waist in the vertical direction during a second period of the foothold period, and time-series data of the angle of the ankle joint during a third period of the foothold period are detected,

in the determining, the fall risk of the test subject is determined using an average value of the time-series data of the displacement of the waist in the vertical direction in the first period, an average value of the time-series data of the displacement of the waist in the vertical direction in the second period, and an average value of the time-series data of the angle of the ankle joint in the third period.

15. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the angle of the knee joint during a first period of the stance phase and time-series data of the angle of the ankle joint during a second period of the foot suspension phase are detected,

in the determining, the fall risk of the test subject is determined using an average value of the time-series data of the angle of the knee joint in the first period and an average value of the time-series data of the angle of the ankle joint in the second period.

16. Fall risk assessment method according to claim 1,

in the detecting, time-series data of the displacement of the waist in the vertical direction during a first period of the stance phase, time-series data of the angle of the knee joint during a second period of the stance phase, and time-series data of the angle of the ankle joint during a third period of the foot suspension phase are detected,

in the determining, the fall risk of the test subject is determined using an average value of the time-series data of the displacement of the waist in the vertical direction in the first period, an average value of the time-series data of the angle of the knee joint in the second period, and an average value of the time-series data of the angle of the ankle joint in the third period.

17. Fall risk assessment method according to any of claims 1 to 16,

in the determining, it is determined that the subject has the risk of falling when the displacement of the waist in the vertical direction in the stance phase is smaller than a threshold value, when the displacement of the waist in the vertical direction in the foot suspension phase is smaller than a threshold value, when the angle of the knee joint in the stance phase is smaller than a threshold value, or when the angle of the ankle joint in the foot suspension phase is smaller than a threshold value.

18. Fall risk assessment method according to any of claims 1 to 16,

in the determining, whether the subject has the risk of falling is determined by inputting at least one of the detected displacement of the waist in the vertical direction during the footfall period, the angle of the knee joint in the footfall period, and the angle of the ankle joint in the footfall period into a prediction model, wherein the prediction model generates at least one of the displacement of the waist in the vertical direction during the footfall period, the angle of the knee joint in the footfall period, and the angle of the ankle joint in the footfall period as an input value, and the risk of falling is generated as an output value.

19. A fall risk assessment device for assessing fall risk based on walking motion of a test subject, comprising:

an acquisition unit configured to acquire walking data relating to walking of the test subject;

a detection unit configured to detect, based on the walking data, at least one of a vertical displacement of the waist of the test subject in a stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject in a foot suspension phase of the one leg, an angle of a knee joint of the one leg in the stance phase, and an angle of an ankle joint of the one leg in the foot suspension phase; and the number of the first and second groups,

and a determination unit configured to determine a risk of falling of the test subject using at least one of the displacement of the waist in the vertical direction during the stance phase, the displacement of the waist in the vertical direction during the foot suspension phase, the angle of the knee joint during the stance phase, and the angle of the ankle joint during the foot suspension phase.

20. A non-transitory computer-readable recording medium storing a fall risk assessment program for assessing a fall risk based on a walking motion of a subject, the fall risk assessment program causing a computer to function as:

acquiring walking data related to the walking of the test subject;

Technical Field

The present invention relates to a technique for evaluating a fall risk based on a walking motion of a test subject.

Background

In recent years, techniques have been developed to easily estimate the physical functions of the elderly in order to understand their health status. In particular, the elderly have a high possibility of falling down due to a low body function, and may be in a state of fracture or bedridden due to falling down. Therefore, it is necessary to find an elderly person who is likely to fall, that is, an elderly person who is at risk of falling, as soon as possible, and take measures to prevent falling.

Conventionally, a technique for evaluating cognitive function or motor function based on a parameter measured from walking performed on a daily basis has been proposed.

For example, japanese patent laid-open publication No. 2013-255786 discloses a method of evaluating the degree of susceptibility to the occurrence of an elderly disorder (risk of an elderly disorder) based on walking parameters measured in walking behavior.

Further, for example, japanese patent laid-open publication No. 2018-114319 proposes a technique of measuring a front-rear acceleration, a left-right acceleration, and a vertical acceleration of a test subject during movement by an acceleration sensor attached to a waist portion of the test subject, and evaluating a movement capability based on changes in the front-rear acceleration, the left-right acceleration, and the vertical acceleration with time.

However, in the above-described conventional techniques, it is difficult to evaluate the fall risk simply and with high accuracy, and further improvement is required.

Disclosure of Invention

The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a technique for easily and accurately estimating a fall risk.

A fall risk assessment method according to an aspect of the present invention is a fall risk assessment method for a fall risk assessment apparatus for assessing a fall risk based on a walking motion of a test subject, including the steps of: acquiring walking data related to the walking of the test subject; detecting at least one of a vertical displacement of the waist of the test subject during the stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject during the foot suspension phase of the one leg, an angle of the knee joint of the one leg during the stance phase, and an angle of the ankle joint of the one leg during the foot suspension phase, based on the walking data; and determining the falling risk of the test object by using at least one of the displacement of the waist in the vertical direction in the standing period, the displacement of the waist in the vertical direction in the foot suspension period, the angle of the knee joint in the standing period and the angle of the ankle joint in the foot suspension period.

Drawings

Fig. 1 is a block diagram showing a configuration of a fall risk assessment system according to an embodiment of the present invention.

Fig. 2 is a schematic diagram for explaining the process of extracting bone data from two-dimensional image data according to the present embodiment.

Fig. 3 is a schematic diagram for explaining a walking cycle according to the present embodiment.

Fig. 4 is a flowchart for explaining fall risk assessment processing using walking motion of a test subject in the present embodiment.

Fig. 5 is a flowchart for explaining the fall risk determination process of step S4 of fig. 4.

Fig. 6 is a flowchart for explaining another example of the fall risk determination processing in step S4 in fig. 4.

Fig. 7 is a schematic diagram showing changes in the vertical displacement of the waist in one walking cycle in the present embodiment.

Fig. 8 is a schematic diagram showing an ROC curve obtained from the result of determining whether or not there is a risk of falling using the prediction model of the present embodiment.

Fig. 9 is a schematic diagram showing an ROC curve obtained from the result of determining whether there is a risk of falling using the prediction model according to the first modification of the present embodiment.

Fig. 10 is a schematic diagram showing an average of time-series data of vertical displacements of the waist of subjects who do not have a risk of falling during 9% to 19% of one walking cycle and an average of time-series data of vertical displacements of the waist of subjects who do not have a risk of falling during 9% to 19% of one walking cycle in the first modification of the present embodiment.

Fig. 11 is a schematic diagram showing an ROC curve obtained from the result of determining whether there is a risk of falling using the prediction model according to the second modification of the present embodiment.

Fig. 12 is a schematic diagram showing a change in angle of one knee joint in one walking cycle in a third modification of the present embodiment.

Fig. 13 is a schematic diagram showing an ROC curve obtained from the result of determining whether there is a fall risk using the prediction model according to the third modification of the present embodiment.

Fig. 14 is a schematic diagram showing an ROC curve obtained from the result of determining the presence or absence of a fall risk using the prediction model according to the fourth modification of the present embodiment.

Fig. 15 is a schematic diagram showing the average of the angles of the knee joints of the test subjects having no risk of falling at 35% of the time of one walking cycle of one leg and the average of the angles of the knee joints of the test subjects having a risk of falling at 35% of the time of one walking cycle of one leg in the fourth modification of the present embodiment.

Fig. 16 is a schematic diagram showing a change in angle of one ankle joint in one walking cycle in a fifth modification of the present embodiment.

Fig. 17 is a schematic diagram showing an ROC curve obtained from the result of determining whether or not there is a risk of falling using the prediction model according to the fifth modification of the present embodiment.

Fig. 18 is a schematic diagram showing an ROC curve obtained from the result of determining whether there is a risk of falling using the prediction model according to the sixth modification of the present embodiment.

Fig. 19 is a diagram showing, in a sixth modification of the present embodiment, the average of the average values of the time-series data of the angles of the ankle joint of one leg of the subjects who do not have the risk of falling during the period of 84% to 89% of one walking cycle and the average value of the time-series data of the angles of the ankle joint of one leg of the subjects who have the risk of falling during the period of 84% to 89% of one walking cycle.

Fig. 20 is a schematic diagram showing an ROC curve obtained from the result of determining the presence or absence of a fall risk using the prediction model according to the seventh modification of the present embodiment.

Fig. 21 is a schematic diagram showing an ROC curve obtained from the result of determining whether there is a risk of falling using the prediction model according to the eighth modification of the present embodiment.

Fig. 22 is a schematic diagram showing an ROC curve obtained from the result of determining the presence or absence of a fall risk using the prediction model according to the ninth modification of the present embodiment.

Fig. 23 is a schematic diagram showing an ROC curve obtained from the result of determining whether there is a fall risk using the prediction model according to the tenth modification of the present embodiment.

Fig. 24 is a schematic diagram showing an ROC curve obtained from the result of determining the presence or absence of a fall risk using the prediction model according to the eleventh modification of the present embodiment.

Fig. 25 is a schematic diagram showing an example of the evaluation result screen displayed in the present embodiment.

Detailed Description

(basic knowledge of the invention)

The measurement of walking parameters disclosed in japanese patent laid-open publication No. 2013 and 255786 uses a sheet-type pressure sensor or a three-dimensional motion analysis system. The sheet type pressure sensor measures pressure distribution during walking and measures walking parameters based on the pressure distribution. The three-dimensional motion analysis system acquires image information, which is obtained by photographing a tag mounted on a leg, from a plurality of cameras, and measures a walking parameter by analyzing a motion from the image information. The arrangement of such a sheet-type pressure sensor or three-dimensional motion analysis system is troublesome and laborious. For this reason, it is difficult to easily evaluate the risk of the aged disorder with the technique disclosed in japanese patent laid-open publication No. 2013-255786.

Further, as the walking parameters of the technique disclosed in japanese patent laid-open publication No. 2013-255786, two or more walking parameters selected from the speed of riding a bicycle, the stride, the walking ratio, the stride length, the step interval, the walking angle, the toe angle, the left-right difference of the stride, the left-right difference of the step interval, the left-right difference of the walking angle, and the left-right difference of the biped support period are used. The walking angle is an angle formed between a line connecting the heel of the left and right legs and the heel of the other leg and the traveling direction. The toe angle is the angle formed between the line connecting the heel and the toe and the direction of travel. Further, the technique disclosed in japanese patent laid-open publication No. 2013-255786 evaluates the risk of an aged disorder selected from at least knee pain, lumbago, urinary incontinence, dementia, and sarcopenia. However, japanese patent laid-open publication No. 2013-255786 does not disclose the evaluation of the risk of the elderly disorder using walking parameters other than the above, and the accuracy of the evaluation of the risk of the elderly disorder can be further improved by using other walking parameters.

The mobility evaluation device disclosed in japanese patent laid-open publication No. 2018-114319 evaluates at least one of the front-back balance, the weight shift, and the left-right balance of the test subject during movement, based on the front-back acceleration, the left-right acceleration, and the up-down acceleration of the test subject during movement. However, japanese patent laid-open publication No. 2018-114319 does not disclose that the fall risk is estimated using parameters other than those described above, and the accuracy of the fall risk estimation can be further improved by using other walking parameters.

In order to solve the above problems, a fall risk assessment method according to an aspect of the present invention is a fall risk assessment method for a fall risk assessment apparatus for assessing a fall risk based on a walking motion of a test subject, including the steps of: acquiring walking data related to the walking of the test subject; detecting at least one of a vertical displacement of the waist of the test subject during the stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject during the foot suspension phase of the one leg, an angle of the knee joint of the one leg during the stance phase, and an angle of the ankle joint of the one leg during the foot suspension phase, based on the walking data; and determining the falling risk of the test object by using at least one of the displacement of the waist in the vertical direction in the standing period, the displacement of the waist in the vertical direction in the foot suspension period, the angle of the knee joint in the standing period and the angle of the ankle joint in the foot suspension period.

According to this configuration, at least one of the vertical displacement of the waist of the subject who is walking during the stance phase of one leg, the vertical displacement of the waist during the foot suspension phase of one leg, the angle of the knee joint of one leg during the stance phase, and the angle of the ankle joint of one leg during the foot suspension phase is used as the parameter relating to the fall risk of the subject. The walking motion of the test subject at risk of falling is different from the walking motion of the test subject without risk of falling. Therefore, since the presence or absence of the risk of falling of the test subject is determined by using the parameter relating to the risk of falling of the test subject on foot, the risk of falling of the test subject can be evaluated with high accuracy.

Further, at least one of the vertical displacement of the waist of the subject who is walking in the stance phase of one leg, the vertical displacement of the waist in the foot suspension phase of one leg, the angle of the knee joint of one leg in the stance phase, and the angle of the ankle joint of one leg in the foot suspension phase can be easily detected from, for example, image data obtained by imaging the subject who is walking, and therefore, a large-scale apparatus is not required. For this reason, the present configuration can simply evaluate the fall risk of the test subject.

In the above fall risk assessment method, the detection may detect time-series data of the displacement of the waist in the vertical direction during a predetermined period of the stance phase, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the displacement of the waist in the vertical direction.

The displacement of the waist of the subject who is walking in the vertical direction during the prescribed period of the stance phase of one leg is clearly different between subjects who are not at risk of falling and subjects who are at risk of falling. Therefore, according to this configuration, the risk of falling of the test subject can be reliably evaluated by using the average value of the time-series data of the vertical displacement of the waist of the test subject who is walking during the predetermined period of the stance phase of one leg.

In the above fall risk assessment method, when a period from when one foot of the test subject lands on the ground to when the other foot lands on the ground is expressed as one walking cycle and the one walking cycle is expressed by 1% to 100%, the predetermined period may be 1% to 60% of the one walking cycle.

According to this configuration, the period from the landing of one foot of the test subject to the landing of the other foot is expressed as one walking cycle, and one walking cycle is expressed by 1% to 100%. At this time, by using the average value of the time-series data of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the fall risk of the test subject can be reliably evaluated.

According to this configuration, the fall risk of the test subject can be estimated more reliably by using the average value of the time-series data of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle.

In the above fall risk assessment method, the detection may detect time-series data of the displacement of the waist in the vertical direction during a predetermined period of the foot suspension period, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the displacement of the waist in the vertical direction.

The displacement of the waist of a walking test subject in the vertical direction during the specified period of the foot suspension period of one leg is clearly different between test subjects at risk of falling and test subjects without risk of falling. Therefore, according to this configuration, the risk of falling of the test subject can be reliably evaluated by using the average value of the time-series data of the vertical displacement of the waist of the test subject who is walking during the predetermined period of the foot suspension period of one leg.

According to this configuration, the period from the landing of one foot of the test subject to the landing of the other foot is expressed as one walking cycle, and one walking cycle is expressed by 1% to 100%. At this time, by using the average value of the time-series data of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle, the fall risk of the test subject can be reliably evaluated.

In the above fall risk assessment method, the detection may detect time-series data of an angle of the knee joint during a predetermined period of the stance phase, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the angle of the knee joint.

The angle of the knee joint of the subject who is walking during the prescribed period of the stance phase of one leg is significantly different between subjects who are at risk of falling and subjects who are not at risk of falling. Therefore, according to this configuration, the risk of falling of the test subject can be reliably evaluated by using the average value of the time-series data of the angle of the knee joint of the test subject who is walking during a predetermined period of the stance phase of one leg.

According to this configuration, the period from the landing of one foot of the test subject to the landing of the other foot is expressed as one walking cycle, and one walking cycle is expressed by 1% to 100%. At this time, by using the average value of the time-series data of the angle of the knee joint during 1% to 60% of one walking cycle, the fall risk of the test subject can be reliably evaluated.

In the above fall risk assessment method, when a period from when one foot of the test subject lands on the ground to when the one foot again lands on the ground is expressed as one walking cycle and the one walking cycle is expressed by 1% to 100%, the fall risk of the test subject may be determined by using the angle at which the knee joint is 35% of the time of the one walking cycle in the determination.

According to this configuration, the risk of falling of the test subject can be evaluated more reliably by using the angle of the knee joint at 35% of the time of one walking cycle.

In the above fall risk assessment method, the detection may detect time-series data of the angle of the ankle joint during a predetermined period of the foot suspension period, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the angle of the ankle joint.

The angle of the ankle joint of a walking test subject during the prescribed period of the foot suspension period for one leg is clearly different between test subjects at risk of falling and test subjects without risk of falling. Therefore, according to this configuration, the risk of falling of the test subject can be reliably evaluated by using the average value of the time-series data of the angle of the ankle joint of the test subject who is walking during a predetermined period of the foot suspension period of one leg.

According to this configuration, the period from the landing of one foot of the test subject to the landing of the other foot is expressed as one walking cycle, and one walking cycle is expressed by 1% to 100%. At this time, the risk of falling of the test subject can be reliably evaluated by using the average value of the time-series data of the angle of the ankle joint during 61% to 100% of one walking cycle.

According to this configuration, the risk of falling of the test subject can be estimated more reliably by using the average value of the time-series data of the angle of the ankle joint during 84% to 89% of one walking cycle.

In the above fall risk assessment method, the detection may detect time-series data of the displacement of the waist in the vertical direction during a first period of the stance phase and time-series data of the angle of the knee joint during a second period of the stance phase, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the displacement of the waist in the vertical direction during the first period and an average value of the time-series data of the angle of the knee joint during the second period.

According to this configuration, by combining the average value of the time-series data using the displacement of the waist in the vertical direction during the first period of the stance phase of one leg with the average value of the time-series data using the angle of the knee joint during the second period of the stance phase of one leg, the risk of falling of the test subject can be evaluated with higher accuracy than when the average value of the time-series data using the displacement of the waist in the vertical direction during the first period of the stance phase of one leg or the average value of the time-series data using the angle of the knee joint during the second period of the stance phase of one leg are used.

In the above fall risk assessment method, the detection may be performed by detecting time-series data of the displacement of the waist in the vertical direction during a first period of the stance phase, time-series data of the displacement of the waist in the vertical direction during a second period of the foot suspension phase, and time-series data of the angle of the ankle joint during a third period of the foot suspension phase, in the determining, the fall risk of the test subject is determined using an average value of the time-series data of the displacement of the waist in the vertical direction in the first period, an average value of the time-series data of the displacement of the waist in the vertical direction in the second period, and an average value of the time-series data of the angle of the ankle joint in the third period.

According to this configuration, the average value of the time-series data using the displacement of the waist in the vertical direction during the first period of the footfall period of one leg, the average value of the time-series data using the displacement of the waist in the vertical direction during the second period of the footfall period of one leg, and the average value of the time-series data using the angle of the ankle joint during the third period of the footfall period of one leg are combined, and the risk of falling of the test subject can be estimated with higher accuracy than the average value of the time-series data using the displacement of the waist in the vertical direction during the first period of the footfall period of one leg, the average value of the time-series data using the displacement of the waist in the vertical direction during the second period of the footfall period of one leg, and the average value of the time-series data using the angle of the ankle joint during the third period of the footfall period of one leg, respectively.

In the above fall risk assessment method, the detection may detect time-series data of the angle of the knee joint in a first period of the footfall period and time-series data of the angle of the ankle joint in a second period of the foot-floating period, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the angle of the knee joint in the first period and an average value of the time-series data of the angle of the ankle joint in the second period.

According to this configuration, by combining the average value of the time-series data of the angle of the knee joint in the first period of the footfall period of one leg with the average value of the time-series data of the angle of the ankle joint in the second period of the footfall period of one leg, the risk of falling of the test subject can be evaluated with higher accuracy than by using the average value of the time-series data of the angle of the knee joint in the first period of the footfall period of one leg and the average value of the time-series data of the angle of the ankle joint in the second period of the footfall period of one leg, respectively.

In the above fall risk assessment method, the detection may detect time-series data of the displacement of the waist in the vertical direction in the first period of the footfall period, time-series data of the angle of the knee joint in the second period of the footfall period, and time-series data of the angle of the ankle joint in the third period of the foot-floating period, and the determination may determine the fall risk of the test subject using an average value of the time-series data of the displacement of the waist in the vertical direction in the first period, an average value of the time-series data of the angle of the knee joint in the second period, and an average value of the time-series data of the angle of the ankle joint in the third period.

According to this configuration, by combining the average value of the time-series data of the displacement of the waist in the vertical direction in the first period of the stance phase of one leg, the average value of the time-series data of the angle of the knee joint in the second period of the stance phase of one leg, and the average value of the time-series data of the angle of the ankle joint in the third period of the foot suspension phase of one leg, the risk of falling of the test subject can be evaluated with higher accuracy than the average value of the time-series data of the displacement of the waist in the vertical direction in the first period of the stance phase of one leg, the average value of the time-series data of the angle of the knee joint in the second period of the stance phase of one leg, and the average value of the time-series data of the angle of the ankle joint in the third period of the foot suspension phase of one leg, respectively.

In the above fall risk assessment method, the determination may be made that the subject is at risk of falling when the displacement of the waist in the vertical direction during the stance phase is less than a threshold value, when the displacement of the waist in the vertical direction during the foot suspension phase is less than a threshold value, when the angle of the knee joint during the stance phase is less than a threshold value, or when the angle of the ankle joint during the foot suspension phase is less than a threshold value.

According to this configuration, it is determined that the test subject is at risk of falling when the vertical displacement of the waist in the stance phase is smaller than the threshold, when the vertical displacement of the waist in the foot suspension phase is smaller than the threshold, when the angle of the knee joint in the stance phase is smaller than the threshold, or when the angle of the ankle joint in the foot suspension phase is smaller than the threshold. Therefore, whether the test subject has a risk of falling can be easily determined by comparing the displacement of the waist in the vertical direction during the standing period, the displacement of the waist in the vertical direction during the foot suspension period, the angle of the knee joint during the standing period, and the angle of the ankle joint during the foot suspension period with the threshold values.

In the above fall risk assessment method, the determination may be made by inputting at least one of the detected displacement of the waist in the vertical direction during the stance phase, the detected displacement of the waist in the vertical direction during the foot suspension phase, the detected angle of the knee joint during the stance phase, and the detected angle of the ankle joint during the foot suspension phase into a prediction model to determine whether the subject is at risk of falling, wherein the prediction model is generated with at least one of the displacement of the waist in the vertical direction during the stance phase, the displacement of the waist in the vertical direction during the foot suspension phase, the angle of the knee joint during the stance phase, and the angle of the ankle joint during the foot suspension phase as an input value, and with the presence or absence of the fall risk of the test subject as an output value.

According to this configuration, the prediction model is generated using at least one of the vertical displacement of the waist in the stance phase, the vertical displacement of the waist in the foot suspension phase, the angle of the knee joint in the stance phase, and the angle of the ankle joint in the foot suspension phase as an input value, and using as an output value whether or not the test subject is at risk of falling. Then, at least one of the detected displacement of the waist in the vertical direction during the stance phase, the displacement of the waist in the vertical direction during the foot suspension phase, the angle of the knee joint during the stance phase, and the angle of the ankle joint during the foot suspension phase is input to the prediction model, thereby determining whether the test subject is at risk of falling. Therefore, by storing the prediction model in advance, it is possible to easily determine whether or not the test subject is at risk of falling.

A fall risk assessment device according to another aspect of the present invention is a fall risk assessment device for assessing a fall risk based on a walking motion of a test subject, including: an acquisition unit configured to acquire walking data relating to walking of the test subject; a detection unit configured to detect, based on the walking data, at least one of a vertical displacement of the waist of the test subject in a stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject in a foot suspension phase of the one leg, an angle of a knee joint of the one leg in the stance phase, and an angle of an ankle joint of the one leg in the foot suspension phase; and a determination unit configured to determine a risk of falling of the test subject using at least one of the displacement of the waist portion in the vertical direction during the stance phase, the displacement of the waist portion in the vertical direction during the foot suspension phase, the angle of the knee joint during the stance phase, and the angle of the ankle joint during the foot suspension phase.

A recording medium according to another aspect of the present invention is a non-transitory computer-readable recording medium recording a fall risk assessment program for assessing a fall risk based on a walking motion of a test subject, the fall risk assessment program having a function of acquiring walking data relating to the walking of the test subject; detecting at least one of a vertical displacement of the waist of the test subject during the stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject during the foot suspension phase of the one leg, an angle of the knee joint of the one leg during the stance phase, and an angle of the ankle joint of the one leg during the foot suspension phase, based on the walking data; and determining the falling risk of the test object by using at least one of the displacement of the waist in the vertical direction in the standing period, the displacement of the waist in the vertical direction in the foot suspension period, the angle of the knee joint in the standing period and the angle of the ankle joint in the foot suspension period.

Embodiments of the present invention will be described below with reference to the drawings. The embodiments described below are specific examples of the present invention, and are not intended to limit the technical scope of the present invention.

(embodiment mode)

Hereinafter, a fall risk assessment system according to the present embodiment will be described with reference to fig. 1.

Fig. 1 is a block diagram showing a configuration of a fall risk assessment system according to an embodiment of the present invention.

The fall risk assessment system shown in fig. 1 includes a fall risk assessment device 1, a camera 2, and a display unit 3.

The camera 2 photographs a walking test subject. The camera 2 outputs moving image data representing the walking test subject to the fall risk assessment apparatus 1. The camera 2 is connected to the fall risk assessment apparatus 1 by wire or wirelessly.

The fall risk assessment device 1 includes a processor 11 and a memory 12.

The processor 11 is, for example, a CPU (central processing unit), and includes a data acquisition unit 111, a walking parameter detection unit 112, a fall risk determination unit 113, and an evaluation result presentation unit 114.

The memory 12 is a storage device such as a ram (random Access memory), a hdd (hard Disk drive), a ssd (solid State drive), or a flash memory, which can store various information.

The data acquisition unit 111 acquires walking data relating to the walking of the test subject. The walking data is, for example, moving image data obtained by imaging a test subject walking. The data acquisition unit 111 acquires moving image data output by the camera 2.

The walking parameter detection unit 112 extracts bone data representing the bone of the test subject from the moving image data acquired by the data acquisition unit 111. The bone data is represented by coordinates of a plurality of characteristic points representing joints of the test object and the like and straight lines connecting the characteristic points. The walking parameter detection unit 112 may use software (e.g., opendose or 3D dose-base) that detects coordinates of a feature point of a person from two-dimensional image data.

Here, a process of extracting bone data from two-dimensional image data will be described.

Fig. 2 is a schematic diagram for explaining the process of extracting bone data from two-dimensional image data according to the present embodiment.

The walking parameter detection unit 112 extracts the skeleton data 21 from the two-dimensional image data 20 including the image of the test subject 200 that is walking. The skeleton data 21 includes a feature point 201 representing the head, a feature point 202 representing the center of the shoulders, a feature point 203 representing the right shoulder, a feature point 204 representing the right elbow, a feature point 205 representing the right hand, a feature point 206 representing the left shoulder, a feature point 207 representing the left elbow, a feature point 208 representing the left hand, a feature point 209 representing the waist, a feature point 210 representing the right femoral joint, a feature point 211 representing the right knee joint, a feature point 212 representing the right ankle joint, a feature point 213 representing the right toe portion, a feature point 214 representing the left femoral joint, a feature point 215 representing the left knee joint, a feature point 216 representing the left ankle joint, and a feature point 217 representing the left toe portion.

The moving image data is composed of a plurality of two-dimensional image data. The walking parameter detection unit 112 extracts time-series skeleton data from each of a plurality of two-dimensional image data constituting moving image data. The walking parameter detection unit 112 may extract the skeleton data from the two-dimensional image data of all frames, or may extract the skeleton data from the two-dimensional image data of each predetermined frame. In the present embodiment, the fall risk is estimated based mainly on the movement of the lower limbs of the subject during walking. Therefore, the walking parameter detecting unit 112 may extract only the skeletal data of the lower limb of the test subject.

Then, the walking parameter detection unit 112 extracts the time-series skeleton data extracted from the dynamic image data, and cuts out the skeleton data corresponding to one walking cycle of the test subject. The walking motion of a person is a periodic motion.

Here, the walking cycle of the test subject will be described.

Fig. 3 is a schematic diagram for explaining a walking cycle according to the present embodiment.

As shown in fig. 3, the period from when one foot of the test subject lands on the ground to when the foot lands on the ground again is shown as one walking cycle. One walking cycle shown in fig. 3 is a period from when the right foot of the test subject lands on the ground to when the right foot again lands on the ground. Also, one walking cycle is normalized to 1% to 100%. The period of 1% to 60% of a walking cycle is called the stance period in which one foot (e.g., the right foot) falls to the ground, and the period of 61% to 100% of a walking cycle is called the foot suspension period in which one foot (e.g., the right foot) leaves the ground. One gait cycle includes a stance phase and a foot suspension phase. One walking cycle may be a period from the landing of the left foot of the test subject to the landing of the left foot again.

The walking parameter detecting unit 112 detects, based on the walking data, at least one of a vertical displacement of the waist of the test subject in the stance phase of one leg of the test subject, a vertical displacement of the waist of the test subject in the suspension phase of one leg, an angle of the knee joint of one leg in the stance phase, and an angle of the ankle joint of one leg in the suspension phase.

In the present embodiment, the walking parameter detecting unit 112 detects, based on the walking data, the vertical displacement of the waist of the test subject during the stance phase of one leg of the test subject. The walking parameter detecting unit 112 detects a vertical displacement of the waist of the test subject in the stance phase of one leg of the test subject from the time-series skeleton data corresponding to one intercepted walking cycle. As shown in fig. 2, the displacement α of the waist in the vertical direction is the displacement of the characteristic point 209 representing the waist in the vertical direction.

In particular, the walking parameter detecting unit 112 detects time-series data of the vertical displacement of the waist during a predetermined period of the stance phase of one leg. Specifically, the predetermined period is a period of 1% to 60% of one walking cycle. The predetermined period may be a period of 9% to 19% of one walking cycle. The walking parameter detection unit 112 calculates an average value of time-series data of vertical displacement of the waist during a predetermined period of the stance phase of one leg as a walking parameter.

Further, a modified example of the present embodiment will be described with respect to detection of a displacement of the waist of the test subject in the vertical direction during the foot suspension period of one leg of the test subject, an angle of the knee joint of one leg of the test subject during the foot standing period, and an angle of the ankle joint of one leg during the foot suspension period.

The fall risk determination unit 113 determines the risk of falling of the test subject by using at least one of the vertical displacement of the waist in the stance phase, the vertical displacement of the waist in the foot suspension phase, the angle of the knee joint in the stance phase, and the angle of the ankle joint in the foot suspension phase.

In the present embodiment, the fall risk determination unit 113 determines the fall risk of the test subject using the average value of the time-series data of the vertical displacement of the waist.

The fall risk determination unit 113 determines whether or not the test subject is at risk of falling by inputting at least one of the detected displacement of the waist in the vertical direction during the stance phase, the displacement of the waist in the vertical direction during the foot suspension phase, the angle of the knee joint during the stance phase, and the angle of the ankle joint during the foot suspension phase into a prediction model that is generated by using at least one of the displacement of the waist in the vertical direction during the stance phase, the displacement of the waist in the vertical direction during the foot suspension phase, the angle of the knee joint during the stance phase, and the angle of the ankle joint during the foot suspension phase as an input value and using as an output value the presence or absence of risk of falling of the test subject.

In the present embodiment, the fall risk determination unit 113 determines whether or not the subject is at risk of falling by inputting the vertical displacement of the waist detected by the walking parameter detection unit 112 to a prediction model generated with the vertical displacement of the waist as an input value and the vertical displacement of the waist as an output value.

Further, a modified example of the present embodiment will be described with respect to the determination of the risk of falling of the test subject using the displacement of the waist in the vertical direction during the foot suspension period, the angle of the knee joint during the foot standing period, and the angle of the ankle joint during the foot suspension period.

The memory 12 stores in advance a prediction model generated by using the vertical displacement of the waist as an input value and using the fall risk of the test subject as an output value. The prediction model is a regression model in which whether or not the test subject is at risk of falling is used as a target variable, and time-series data of displacement of the waist in the vertical direction in the stance phase of one walking cycle is used as an explanatory variable. The predictive model outputs either a value (e.g., 1) indicating that the subject is at risk of falling or a value (e.g., 0) indicating that the subject is not at risk of falling.

In particular, the fall risk determination unit 113 determines the fall risk of the test subject using the average value of the time-series data of the vertical displacement of the waist in the stance phase of one leg. Specifically, the fall risk determination unit 113 determines the fall risk of the test subject using the average value of the time-series data of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle. The fall risk determination unit 113 may determine the fall risk of the test subject by using an average value of time-series data of vertical displacement of the waist during 9% to 19% of one walking cycle.

In addition, the prediction model may also be generated by machine learning. Examples of machine learning include teacher learning in which a relationship between input and output is learned using teacher data to which a label (output information) is given to input information, teakless learning in which no label is constructed only from input construction data, semi-teacher learning in which a label may be present or absent, and reinforcement learning in which an action of maximizing reward is learned by trial and error. Further, as a specific method of machine learning, there are, for example, a neural network (including deep learning using a multilayer neural network), a genetic algorithm, a decision tree, a bayesian network, a Support Vector Machine (SVM), or the like. In the machine learning of the present invention, any one of the above-listed specific examples may be used.

Furthermore, the predictive model may also output a value representing the degree of fall risk. The value representing the degree of fall risk is, for example, represented by 0.0 to 1.0. In this case, for example, the fall risk determination unit 113 may determine that there is no fall risk when the value indicating the degree of the fall risk is 0.5 or less, and determine that there is a high possibility of the fall risk when the value indicating the degree of the fall risk is greater than 0.5.

The evaluation result presentation unit 114 presents the evaluation result of the fall risk determined by the fall risk determination unit 113. The evaluation result presentation unit 114 outputs the evaluation result determined by the fall risk determination unit 113 to the display unit 3. The evaluation result is at least one of information indicating whether or not the test subject determined by the fall risk determination unit 113 has a fall risk, and an evaluation message.

The display unit 3 displays the evaluation result output from the evaluation result display unit 114. The display unit 3 is, for example, a liquid crystal display panel or a light emitting element.

In addition, the display unit 3 may be configured to show a graph showing the transition of the value showing the degree of the fall risk in order to compare the value showing the degree of the fall risk determined this time with the value showing the degree of the fall risk in the past. In addition, a value indicating the degree of the risk of falling in the past is stored in the memory 12 and can be read out from the memory 12.

The fall risk assessment device 1 may also include a camera 2 and a display unit 3. The fall risk assessment device 1 may also include the display unit 3. The fall risk assessment apparatus 1 may also be a personal computer or a server.

Next, the fall risk assessment processing according to the present embodiment will be described with reference to fig. 4.

Fig. 4 is a flowchart for explaining fall risk assessment processing using walking motion of a test subject in the present embodiment. Fig. 4 shows a flowchart illustrating a sequence of fall risk assessment using the fall risk assessment apparatus 1.

The test subject walks in front of the camera 2. The camera 2 photographs a test subject who is walking. The camera 2 transmits moving image data of the test subject walking to the fall risk assessment apparatus 1.

First, in step S1, the data acquisition unit 111 acquires moving image data transmitted by the camera 2.

Next, in step S2, the walking parameter detecting unit 112 extracts time-series skeleton data from the dynamic image data.

Next, in step S3, the walking parameter detecting unit 112 detects walking parameters for determining the fall risk from the time-series skeleton data. Here, the walking parameter in the present embodiment is an average value of time-series data of displacement of the waist of the test subject in the vertical direction during a predetermined period of the stance phase of one walking cycle. The predetermined period is, for example, a period of 1% to 60% of one walking cycle. The method of determining the walking parameters will be described later.

Next, in step S4, the fall risk determination unit 113 executes fall risk determination processing for determining the fall risk of the test subject using the walking parameters. The fall risk determination process will be described later.

Next, in step S5, the evaluation result presentation unit 114 outputs the evaluation result of the fall risk determined by the fall risk determination unit 113 to the display unit 3. The evaluation result of the fall risk indicates whether the test subject has a fall risk. The evaluation result presentation unit 114 may output, to the display unit 3, not only the presence or absence of the fall risk, but also an evaluation message corresponding to the presence or absence of the fall risk. The display unit 3 displays the evaluation result of the fall risk output from the evaluation result presentation unit 114.

Here, the fall risk determination processing in step S4 in fig. 4 will be described.

Fig. 5 is a flowchart for explaining the fall risk determination process of step S4 of fig. 4.

First, in step S11, the fall risk determination unit 113 reads out the prediction model from the memory 12.

Next, in step S12, the fall risk determination unit 113 inputs the walking parameters detected by the walking parameter detection unit 112 to the prediction model. The walking parameter in the present embodiment is an average value of time-series data of the vertical displacement of the waist of the test subject in a period of 1% to 60% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the vertical displacement of the waist of the test subject in the period of 1% to 60% of one walking cycle to the prediction model.

Next, in step S13, the fall risk determination unit 113 acquires the determination result of the fall risk from the prediction model. The fall risk determination unit 113 obtains whether or not the test subject has a fall risk from the prediction model as a determination result.

In the fall risk determination process according to the present embodiment, the presence or absence of a fall risk is determined by inputting walking parameters to a prediction model generated in advance, but the present invention is not limited to this. As another example of the fall risk determination processing according to the present embodiment, the presence or absence of a fall risk may be determined by comparing a threshold value and a walking parameter stored in advance.

In this case, the memory 12 stores in advance a threshold value for determining whether the test subject is at risk of falling.

The fall risk determination unit 113 may determine that the test subject is at risk of falling when the vertical displacement of the waist in the stance phase is smaller than a threshold value, when the vertical displacement of the waist in the foot suspension phase is smaller than a threshold value, when the angle of the knee joint in the stance phase is smaller than a threshold value, or when the angle of the ankle joint in the foot suspension phase is smaller than a threshold value.

In the present embodiment, the fall risk determination unit 113 may determine that the test subject has a risk of falling when the vertical displacement of the waist is smaller than the threshold value. The fall risk determination unit 113 determines whether or not the average value of the time-series data of the vertical displacement of the waist of the test subject during 1% to 60% of one walking cycle is smaller than a threshold value. The fall risk determination unit 113 determines that the subject is at risk of falling when the average value of the time-series data of the vertical displacement of the waist of the subject during 1% to 60% of one walking cycle is smaller than a threshold value. On the other hand, when the average value of the time-series data of the vertical displacement of the waist of the test subject during 1% to 60% of one walking cycle is equal to or greater than the threshold value, the fall risk determination unit 113 determines that the test subject has no risk of falling, that is, the test subject is a healthy person.

Fig. 6 is a flowchart for explaining another example of the fall risk determination processing in step S4 in fig. 4.

First, in step S21, the fall risk determination unit 113 reads out the threshold value from the memory 12.

Next, in step S22, the fall risk determination unit 113 determines whether or not the walking parameter detected by the walking parameter detection unit 112 is smaller than a threshold value. The walking parameter in the present embodiment is an average value of time-series data of the vertical displacement of the waist of the test subject in a period of 1% to 60% of one walking cycle. The fall risk determination unit 113 determines whether or not the average value of the time-series data of the vertical displacement of the waist of the test subject during 1% to 60% of one walking cycle is smaller than a threshold value.

Here, if it is determined that the walking parameter is smaller than the threshold value (yes in step S22), the fall risk determination unit 113 determines that the test subject has a risk of falling in step S23.

On the other hand, when determining that the walking parameter is equal to or greater than the threshold value (no in step S22), in step S24, the fall risk determination unit 113 determines that the test subject has no risk of falling, that is, the test subject is a healthy subject.

As described above, in the present embodiment, the vertical displacement of the waist of the subject who is walking during the stance phase is a parameter related to the presence or absence of the risk of falling of the subject. The walking motion of the test subject at risk of falling is different from the walking motion of the test subject without risk of falling. Therefore, since the presence or absence of the risk of falling of the test subject is determined by using the parameter relating to the presence or absence of the risk of falling of the test subject during walking, the risk of falling of the test subject can be evaluated with high accuracy.

Further, since the vertical displacement of the waist of the walking test subject during the stance phase can be easily detected from the image data obtained by imaging the walking test subject, for example, a large-sized device is not required. For this reason, the present configuration can simply evaluate the fall risk of the test subject.

The walking parameters and the prediction model according to the present embodiment are determined by experiments. The following describes a method of determining walking parameters and a prediction model according to the present embodiment.

The total number of subjects participating in the experiment was 92. The male test subjects were 27, and the female test subjects were 65. From the results of the past study, the criterion for determining whether or not there is a risk of falling is set to be whether or not the test subject can stand with his or her eyes open for 30 seconds. The holding time was measured in a state where the test subject kept standing with his eyes open. When the holding time is 30 seconds or less, it is determined that there is a risk of falling, and when the holding time is more than 30 seconds, it is determined that there is no risk of falling. As a result of the determination, 34 subjects having a fall risk among the test subjects were determined. Among the test subjects at risk of falling, the male test subjects were 10 persons, and the female test subjects were 24 persons. In the experiment, subjects walked in front of the camera. A test subject is imaged by a camera, and bone data of each test subject is extracted from dynamic image data. Then, time-series data of the displacement of the waist in the vertical direction of each test subject is detected from the extracted skeletal data.

Fig. 7 is a schematic diagram showing changes in the vertical displacement of the waist in one walking cycle in the present embodiment. In fig. 7, the vertical axis represents the displacement of the waist in the vertical direction, and the horizontal axis represents one walking cycle of the normalization. In fig. 7, the broken line shows the average waveform of the vertical displacement of the waist of the test subjects who are not at risk of falling, and the solid line shows the average waveform of the vertical displacement of the waist of the test subjects who are at risk of falling.

In the experiment, one standardized walking cycle was divided into ten sections, and the average value of the displacement of the waist in the vertical direction in one section or two or more continuous sections was calculated for each test subject. Then, a plurality of prediction models are created, each of which has, as an objective variable, whether or not the test subject is at risk of falling, and has, as an explanatory variable, an average value of displacements of the waist in the vertical direction in one or two or more consecutive intervals. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an roc (receiver Operating characteristics) curve is calculated for each of the plurality of prediction models. Then, AUC (area Under curve) values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In the present embodiment, the AUC value of the prediction model created with the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle as an explanatory variable is the highest.

Fig. 8 is a schematic diagram showing an ROC curve obtained from the result of determining whether or not there is a risk of falling using the prediction model of the present embodiment.

The prediction model according to the present embodiment is created with the target variable of whether or not the subject is at risk of falling, and the explanatory variable of the average value of the vertical displacement of the waist during 1% to 60% of one walking cycle. In fig. 8, the vertical axis represents the True Positive Rate (True Positive Rate), and the horizontal axis represents the False Positive Rate (False Positive Rate). The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 8 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created with the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle as an explanatory variable. The AUC value of the ROC curve shown in fig. 8 is 0.733. The AUC value is the area of the lower half of the ROC curve. It is believed that the greater the AUC value (closer to 1) the higher the performance of the prediction model. In this case, the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle is determined as the walking parameter. Then, a prediction model created using as an explanatory variable the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The memory 12 stores in advance a prediction model generated by using as input values an average value of time-series data of displacements of the waist in the vertical direction during 1% to 60% of one walking cycle and as output values whether or not the test subject has a risk of falling. The walking parameter detecting unit 112 detects time-series data of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle to the prediction model, and acquires a determination result indicating whether the test subject has a fall risk from the prediction model.

In addition, in the period of 1% to 60% of one walking cycle shown in fig. 7, the average waveform of the displacement in the vertical direction of the waist of the test subjects with the risk of falling is smaller than the average waveform of the displacement in the vertical direction of the waist of the test subjects without the risk of falling. For this reason, a value between the average of the time-series data of the vertical displacement of the waist of the subject with a fall risk in a period of 1% to 60% of one walking cycle and the average of the time-series data of the vertical displacement of the waist of the subject without a fall risk in a period of 1% to 60% of one walking cycle may be stored in the memory 12 as a threshold value. The fall risk determination unit 113 may determine the fall risk by comparing an average value of time-series data of vertical displacements of the waist of the test subject during 1% to 60% of one walking cycle with a threshold value stored in advance.

In the present embodiment, the walking parameter is an average value of time-series data of displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, but the present invention is not limited to this. Various examples of the walking parameters in the present embodiment will be described below.

First, the walking parameters of the first modification of the present embodiment will be described.

The walking parameter in the first modification of the present embodiment may be an average value of time-series data of vertical displacement of the waist during 9% to 19% of one walking cycle.

In the first modification of the present embodiment, time-series data of displacements of the waist in the vertical direction of each of a plurality of test subjects is detected, as in the above-described experiment. Then, a prediction model is created in which whether the subject has a fall risk or not is used as an objective variable, and an average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle is used as an explanatory variable. The predictive model was evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve of the prediction model is calculated. Then, AUC values of ROC curves of the prediction models are calculated.

The prediction model according to the first modification of the present embodiment is created with the target variable of whether or not the subject is at risk of falling, and the explanatory variable of the average value of the vertical displacement of the waist during 9% to 19% of one walking cycle. In fig. 9, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 9 is a curve that depicts the true positive rate and the false positive rate of a prediction model that takes as an explanatory variable the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle. The AUC value of the ROC curve shown in fig. 9 is 0.8058. In this case, the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle is determined as the walking parameter. Then, a prediction model created using as an explanatory variable the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The memory 12 stores in advance a prediction model generated by using as input values the average of time-series data of vertical displacements of the waist during 9% to 19% of one walking cycle and as output values the presence or absence of a fall risk of the test subject.

The walking parameter detecting unit 112 detects time-series data of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of the vertical displacement of the waist in a period of 9% to 19% of one walking cycle.

The fall risk determination unit 113 determines whether the subject has a risk of falling, using an average value of time-series data of vertical displacement of the waist during 9% to 19% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle to the prediction model, and acquires a determination result indicating whether the test subject has a fall risk from the prediction model.

As shown in fig. 10, the average value of the time-series data of the vertical displacement of the waist of the subjects who do not have the risk of falling during 9% to 19% of one walking cycle was 43.3mm, and the average value of the time-series data of the vertical displacement of the waist of the subjects who have the risk of falling during 9% to 19% of one walking cycle was 34.3 mm.

As described above, the average value of the time-series data of the vertical displacement of the waist of the subjects with the risk of falling is smaller than the average value of the time-series data of the vertical displacement of the waist of the subjects without the risk of falling in the period of 9% to 19% of one walking cycle. For this reason, a value between the average of the time-series data of the vertical displacement of the waist of the subject with a fall risk in 9% to 19% of one walking cycle and the average of the time-series data of the vertical displacement of the waist of the subject without a fall risk in 9% to 19% of one walking cycle may be stored in the memory 12 as a threshold value. The fall risk determination unit 113 may determine whether there is a risk of falling by comparing an average value of time-series data of vertical displacements of the waist of the test subject during 9% to 19% of one walking cycle with a threshold value stored in advance.

Next, walking parameters of a second modification of the present embodiment will be described.

The walking parameter in the second modification of the present embodiment may be an average value of time-series data of the vertical displacement of the waist of the test subject during the foot suspension period of one leg of the test subject.

In the second modification of the present embodiment, as in the above-described experiment, time-series data of vertical displacement of the waist of each of a plurality of subjects is detected from skeletal data of the plurality of subjects including a subject having no risk of falling and a subject having a risk of falling.

In the second modification of the present embodiment, time-series data of displacements of the waist in the vertical direction of each of a plurality of test subjects is detected, as in the above-described experiment. Then, a prediction model is created that uses, as an objective variable, whether the subject is at risk of falling, and uses, as an explanatory variable, an average value of the vertical displacement of the waist during a predetermined period of the foot suspension period. The predetermined period is a period of 61% to 100% of one walking cycle. The predictive model was evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve of the prediction model is calculated. Then, AUC values of ROC curves of the prediction models are calculated.

The prediction model according to the second modification of the present embodiment is created with the target variable of whether or not the test subject is at risk of falling, and the explanatory variable of the average value of the vertical displacement of the waist during 61% to 100% of one walking cycle. In fig. 11, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 11 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created with the average value of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle as an explanatory variable. The AUC value of the ROC curve shown in fig. 11 is 0.713. In this case, the average value of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle is determined as the walking parameter. Then, a prediction model created using as an explanatory variable the average value of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detecting unit 112 detects the vertical displacement of the waist of the test subject based on the walking data. The walking parameter detecting unit 112 detects a vertical displacement of the waist of the test subject from the time-series skeleton data corresponding to the extracted one walking cycle. In particular, the walking parameter detecting unit 112 detects time-series data of the displacement of the waist in the vertical direction during a predetermined period of the foot suspension period of one leg. Specifically, the predetermined period is a period of 61% to 100% of one walking cycle. The walking parameter detecting unit 112 detects time-series data of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of the vertical displacement of the waist in the period of 61% to 100% of one walking cycle.

The memory 12 stores in advance a prediction model generated by using, as input values, the vertical displacement of the waist during a predetermined period of the foot suspension period, and using, as output values, whether or not the test subject has a fall risk. The prediction model is a regression model having, as a target variable, whether or not the subject is at risk of falling, and having, as an explanatory variable, time-series data of displacement of the waist in the vertical direction in one walking cycle. In particular, the memory 12 stores in advance a prediction model generated by using as input values the average of time-series data of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle and as output values the presence or absence of the fall risk of the test subject.

The fall risk determination unit 113 determines whether the test subject is at risk of falling, using an average value of time-series data of vertical displacements of the waist during a predetermined period of the foot suspension period. The fall risk determination unit 113 determines whether or not the subject is at risk of falling by inputting the average value of the time-series data of the vertical displacement of the waist detected by the walking parameter detection unit 112 to a prediction model having the average value of the time-series data of the vertical displacement of the waist during a predetermined period of the foot suspension period as an input value and the fall risk as an output value.

The fall risk determination unit 113 determines whether the test subject is at risk of falling, using an average value of time-series data of the displacement of the waist in the vertical direction during a predetermined period of the foot-suspended period of one leg. Specifically, the fall risk determination unit 113 determines whether the test subject has a risk of falling, using an average value of time-series data of vertical displacement of the waist during 61% to 100% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the displacement of the waist in the vertical direction during 61% to 100% of one walking cycle to the prediction model, and acquires a determination result indicating whether the test subject has a fall risk from the prediction model.

In addition, in the period of 61% to 100% of one walking cycle shown in fig. 7, the average waveform of the displacement in the vertical direction of the waist of the test subjects with the risk of falling is smaller than the average waveform of the displacement in the vertical direction of the waist of the test subjects without the risk of falling. For this reason, a value between the average of the time-series data of the displacement in the vertical direction of the waist of the subject with the fall risk obtained by the experiment during 61% to 100% of one walking cycle and the average of the time-series data of the displacement in the vertical direction of the waist of the subject without the fall risk during 61% to 100% of one walking cycle may be stored in the memory 12 as the threshold value. The fall risk determination unit 113 may determine whether there is a risk of falling by comparing an average value of time-series data of vertical displacements of the waist of the test subject during 61% to 100% of one walking cycle with a threshold value stored in advance.

Next, a walking parameter of a third modification of the present embodiment will be described.

The walking parameter in the third modification of the present embodiment may be an average value of time-series data of angles of the knee joints of one leg during a predetermined period of the stance phase of one leg.

Fig. 12 is a schematic diagram showing a change in angle of one knee joint in one walking cycle in a third modification of the present embodiment. In fig. 12, the vertical axis represents the angle of the knee joint, and the horizontal axis represents one normalized gait cycle. In fig. 12, the broken line shows the average waveform of the angle of one knee joint of the test subjects having no risk of falling, and the solid line shows the average waveform of the angle of one knee joint of the test subjects having a risk of falling.

In the third modification of the present embodiment, as in the above-described experiment, time-series data of the angle of one knee joint of each of a plurality of subjects is detected from bone data of the plurality of subjects including a subject having no risk of falling and a subject having a risk of falling. As shown in fig. 2, the knee joint angle γ is an angle formed between a straight line connecting the feature point 211 representing the right knee joint and the feature point 210 representing the right femoral joint and a straight line connecting the feature point 211 representing the right knee joint and the feature point 212 representing the right ankle joint on an arrow-shaped plane.

In the experiment, one standardized gait cycle was divided into ten intervals, and the average value of the angles of one knee joint in one interval or two or more consecutive intervals was calculated for each test subject. Then, a plurality of prediction models are created, each of which has, as a target variable, whether or not the subject is at risk of falling, and an average value of angles of one knee joint in one interval or two or more consecutive intervals as an explanatory variable. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve is calculated for each of the plurality of prediction models. Then, AUC values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In a third modification of the present embodiment, the AUC value of the prediction model created using the average value of the angles of one knee joint during 1% to 60% of one walking cycle as an explanatory variable is the highest.

The prediction model of the third modification of the present embodiment is created with the target variable of whether or not the test subject is at risk of falling, and the explanatory variable of the average value of the angles of one knee joint in the period of 1% to 60% of one walking cycle. In fig. 13, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 13 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created with the average value of the angles of the knee joint during 1% to 60% of one walking cycle as an explanatory variable. The AUC value of the ROC curve shown in fig. 13 is 0.542. In this case, the average value of the angles of the knee joint during 1% to 60% of one walking cycle is determined as the walking parameter. Then, a prediction model created using as an explanatory variable the average value of the angles of the knee joint during 1% to 60% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detection unit 112 detects the angle of the knee joint of one leg of the test subject from the walking data. The walking parameter detection unit 112 detects the angle of the knee joint of one leg of the test subject from the time-series skeletal data corresponding to one extracted walking cycle. In particular, the walking parameter detecting unit 112 detects time-series data of the angle of the knee joint during a predetermined period of the stance phase of one leg. Specifically, the predetermined period is a period of 1% to 60% of one walking cycle. The walking parameter detection unit 112 detects time-series data of the angle of the knee joint of one leg during 1% to 60% of one walking cycle. The walking parameter detection unit 112 calculates an average value of time-series data of angles of the knee joint of one leg during a period of 1% to 60% of one walking cycle.

In the third modification of the present embodiment, since one walking cycle is a period from the landing of the right foot of the test subject to the landing of the right foot again, the walking parameter detecting unit 112 detects the angle γ of the knee joint of the right leg. When one walking cycle is a period from the landing of the left foot of the test subject to the landing of the left foot again, the walking parameter detecting unit 112 may detect the angle γ of the knee joint of the left leg.

The memory 12 stores in advance a prediction model generated by using the angle of the knee joint as an input value and using whether or not the test subject has a fall risk as an output value. The prediction model is a regression model in which whether or not the subject is at risk of falling is used as a target variable, and time-series data of the angle of the knee joint in one walking cycle is used as an explanatory variable. In particular, the memory 12 stores in advance a prediction model generated by using as input values the average value of time-series data of angles of the knee joint of one leg during a period of 1% to 60% of one walking cycle and using as output values whether or not the test subject has a risk of falling.

The fall risk determination unit 113 determines whether the subject has a risk of falling using the angle of the knee joint. The fall risk determination unit 113 determines whether or not the test subject is at risk of falling by inputting the angle of the knee joint detected by the walking parameter detection unit 112 into a prediction model that is generated with the angle of the knee joint as an input value and the angle of the knee joint as an output value.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using an average value of time-series data of angles of the knee joint in a predetermined period of the stance phase of one leg. Specifically, the fall risk determination unit 113 determines whether the subject is at risk of falling using an average value of time-series data of angles of the knee joint of one leg during a period of 1% to 60% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the angles of the knee joints of one leg during 1% to 60% of one walking cycle to the prediction model, and thereby obtains a determination result indicating whether the test subject has a fall risk from the prediction model.

In addition, in the period of 1% to 60% of one walking cycle shown in fig. 12, the average waveform of the angle of the knee joint of one leg of the test subjects with the risk of falling was smaller than the average waveform of the angle of the knee joint of one leg of the test subjects without the risk of falling. For this reason, a value between the average of the time-series data of the angles of the knee joint of one leg of the subjects with the risk of falling during 1% to 60% of one walking cycle and the average of the time-series data of the angles of the knee joint of one leg of the subjects without the risk of falling during 1% to 60% of one walking cycle, which are obtained through experiments, may be stored in the memory 12 as the threshold value. The fall risk determination unit 113 may determine the presence or absence of a fall risk by comparing an average value of time-series data of angles of the knee joint of one leg of the test subject during 1% to 60% of one walking cycle with a threshold value stored in advance.

Next, a walking parameter of a fourth modification of the present embodiment will be described.

The walking parameter in the fourth modification of the present embodiment may be an angle of the knee joint of one leg at a predetermined time during the stance phase of one leg.

In the fourth modification of the present embodiment, as in the above-described experiment, time-series data of the angle of one knee joint of each of a plurality of subjects is detected from bone data of the plurality of subjects including a subject having no risk of falling and a subject having a risk of falling. In addition, a prediction model is created that has as a target variable whether or not the subject is at risk of falling, and as an explanatory variable an angle of one knee joint at 35% of the time of one walking cycle. The predictive model was evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve of the prediction model is calculated. Then, AUC values of ROC curves of the prediction models are calculated.

The prediction model of the fourth modification of the present embodiment is created with the target variable of whether or not the subject is at risk of falling, and the explanatory variable of the angle of one knee joint at 35% of the time of one walking cycle. In fig. 14, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 14 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created with the angle of the knee joint at 35% of the time of one gait cycle as an explanatory variable. The AUC value of the ROC curve shown in fig. 14 is 0.6242. In this case, the angle of the knee joint at the time of 35% of one walking cycle is determined as the walking parameter. Then, a prediction model created with the angle of the knee joint at 35% of the time of one walking cycle as an explanatory variable is determined as the prediction model used by the fall risk determination unit 113.

The memory 12 stores in advance a prediction model generated by using the angle of the knee joint as an input value and using whether or not the test subject has a fall risk as an output value. The prediction model is a regression model in which whether or not the subject is at risk of falling is used as a target variable, and time-series data of the angle of the knee joint in one walking cycle is used as an explanatory variable. In particular, the memory 12 stores in advance a prediction model generated by using, as input values, angles of the knee joints at 35% of the time of one walking cycle of one leg and using, as output values, whether or not the test subject is at risk of falling.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using an average value of time-series data of angles of the knee joint in a predetermined period of the stance phase of one leg. Specifically, the fall risk determination unit 113 determines whether the test subject is at risk of falling, using the angle of the knee joint at 35% of the time of one walking cycle of one leg. The fall risk determination unit 113 inputs the angle of the knee joint at 35% of the time of one walking cycle of one leg to the prediction model, and acquires the determination result indicating whether or not the test subject has a fall risk from the prediction model.

As shown in fig. 15, the average of the angles of the knee joints of the test subjects with no risk of falling at the time 35% of one walking cycle of one leg was 41.0 degrees, and the average of the angles of the knee joints of the test subjects with a risk of falling at the time 35% of one walking cycle of one leg was 36.6 degrees.

As described above, the average of the angles of the knee joints of one leg of the test subjects with the risk of falling was smaller than the average of the angles of the knee joints of one leg of the test subjects without the risk of falling at the time of 35% of one walking cycle. Therefore, the value between the average of the angles of the knee joints of the test subjects with the risk of falling at 35% of the one walking cycle of one leg and the average of the angles of the knee joints of the test subjects without the risk of falling at 35% of the one walking cycle of one leg, which are obtained through the experiment, may be stored in the memory 12 as the threshold value. The fall risk determination unit 113 may determine the presence or absence of a fall risk by comparing the angle of the knee joint of the test subject at 35% of the time of one walking cycle of one leg with a threshold value stored in advance.

Next, walking parameters of a fifth modification of the present embodiment will be described.

The walking parameter in the fifth modification of the present embodiment may be an average value of time-series data of angles of ankle joints of one leg during a predetermined period of the foot suspension period of one leg.

Fig. 16 is a schematic diagram showing a change in angle of one ankle joint in one walking cycle in a fifth modification of the present embodiment. In fig. 16, the vertical axis represents the angle of the ankle joint, and the horizontal axis represents one standardized walking cycle. In fig. 16, the broken line shows the average waveform of the angle of one ankle joint of the test subjects who are not at risk of falling, and the solid line shows the average waveform of the angle of one ankle joint of the test subjects who are at risk of falling.

In the fifth modification of the present embodiment, as in the above-described experiment, time-series data of the angle of one ankle joint of each of a plurality of subjects is detected from bone data of the plurality of subjects including a subject having no risk of falling and a subject having a risk of falling. As shown in fig. 2, the ankle joint angle θ is an angle formed between a straight line connecting a characteristic point 212 representing the right ankle joint and a characteristic point 211 representing the right knee joint and a straight line connecting the characteristic point 212 representing the right ankle joint and a characteristic point 213 representing the right toe.

In the experiment, one standardized walking cycle was divided into ten sections, and the average value of the angles of one ankle joint in one section or two or more continuous sections was calculated for each test subject. Then, a plurality of prediction models are created, each of which has, as an objective variable, whether or not the subject is at risk of falling, and has, as an explanatory variable, an average value of angles of one ankle joint in one section or two or more consecutive sections. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve is calculated for each of the plurality of prediction models. Then, AUC values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In a fifth modification of the present embodiment, the AUC value of the prediction model created using the average value of the angles of one ankle joint during 61% to 100% of one walking cycle as an explanatory variable is the highest.

The prediction model according to the fifth modification of the present embodiment is created with the target variable of whether or not the subject is at risk of falling, and the explanatory variable of the average value of the angles of one ankle joint in the period of 61% to 100% of one walking cycle. In fig. 17, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 17 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created with the average value of the angles of the ankle joint during 61% to 100% of one walking cycle as an explanatory variable. The AUC value of the ROC curve shown in fig. 17 was 0.595. In this case, the average value of the angle of the ankle joint during 61% to 100% of one walking cycle is determined as the walking parameter. A prediction model created using as an explanatory variable the average value of the angles of the ankle joints in the period of 61% to 100% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detecting unit 112 detects the angle of the ankle joint of one leg of the test subject based on the walking data. The walking parameter detecting unit 112 detects the angle of the ankle joint of one leg of the test subject from the time-series skeleton data corresponding to one intercepted walking cycle. In particular, the walking parameter detecting unit 112 detects time-series data of the angle of the ankle joint in a predetermined period of the foot suspension period of one leg. Specifically, the predetermined period is a period of 61% to 100% of one walking cycle. The walking parameter detecting unit 112 detects time-series data of the angle of the ankle joint of one leg during 61% to 100% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of angles of ankle joints of one leg in a period of 61% to 100% of one walking cycle.

In the fifth modification of the present embodiment, since one walking cycle is a period from landing of the right foot of the test subject to landing of the right foot again, the walking parameter detecting unit 112 detects the angle θ of the ankle joint of the right leg. When one walking cycle is a period from the landing of the left foot of the test subject to the landing of the left foot again, the walking parameter detecting unit 112 may detect the angle θ of the ankle joint of the left leg.

The memory 12 stores in advance a prediction model generated by using the angle of the ankle joint as an input value and using the presence or absence of a fall risk of the test subject as an output value. The prediction model is a regression model in which whether or not the subject is at risk of falling is used as a target variable, and time-series data of the ankle joint in one walking cycle is used as an explanatory variable. In particular, the memory 12 stores in advance a prediction model generated by using as input values the average value of time-series data of angles of ankle joints of one leg during a period of 61% to 100% of one walking cycle and using as output values whether or not the test subject is at risk of falling.

The fall risk determination unit 113 determines whether the test subject has a risk of falling, using the angle of the ankle joint. The fall risk determination unit 113 determines whether or not the subject is at risk of falling by inputting the angle of the ankle joint detected by the walking parameter detection unit 112 into a prediction model that is generated with the angle of the ankle joint as an input value and the angle of the ankle joint as an output value.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using an average value of time-series data of angles of ankle joints in a predetermined period of the foot suspension period of one leg. Specifically, the fall risk determination unit 113 determines whether the subject is at risk of falling, using an average value of time-series data of angles of ankle joints of one leg during a period of 61% to 100% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the angle of the ankle joint of one leg during the period of 61% to 100% of one walking cycle to the prediction model, and acquires a determination result indicating whether or not the test subject is at risk of falling from the prediction model.

In addition, in the period of 61% to 100% of one walking cycle shown in fig. 16, the average waveform of the angle of the ankle joint of one leg of the test subjects with the risk of falling was smaller than the average waveform of the angle of the ankle joint of one leg of the test subjects without the risk of falling. For this purpose, a value between the average of the time-series data of the angle of the ankle joint of one leg of the subject who has a risk of falling during the period of 61% to 100% of one walking cycle and the average of the time-series data of the angle of the ankle joint of one leg of the subject who does not have a risk of falling during the period of 61% to 100% of one walking cycle, which are obtained through experiments, may be stored in the memory 12 as the threshold value. The fall risk determination unit 113 may determine whether there is a risk of falling by comparing an average value of time-series data of angles of the ankle joint of one leg of the test subject during 61% to 100% of one walking cycle with a threshold value stored in advance.

Next, walking parameters of a sixth modification of the present embodiment will be described.

The walking parameter in the sixth modification of the present embodiment may be an average value of time-series data of angles of ankle joints of one leg during 84% to 89% of one walking cycle.

In a sixth modification of the present embodiment, time-series data of the angle of the ankle joint of one leg of each of a plurality of test subjects is detected, as in the above-described experiment. Furthermore, a prediction model was created that had the subject's fall risk as the objective variable and the average of the angles of the ankle joints of one leg during 84% to 89% of one walking cycle as the explanatory variable. The predictive model was evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve of the prediction model is calculated. Then, AUC values of ROC curves of the prediction models are calculated.

The prediction model according to the sixth modification of the present embodiment is created with the target variable of whether or not the subject is at risk of falling, and the explanatory variable of the average value of the angles of one ankle joint in a period of 84% to 89% of one walking cycle. In fig. 18, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 18 is a curve plotting a true positive rate and a false positive rate of a prediction model created using, as explanatory variables, an average value of angles of ankle joints during 84% to 89% of one walking cycle. The AUC value of the ROC curve shown in fig. 18 is 0.5928. In this case, the average value of the angle of the ankle joint during 84% to 89% of one walking cycle is determined as the walking parameter. A prediction model created using as an explanatory variable the average value of the angles of the ankle joints in the period of 84% to 89% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The memory 12 stores in advance a prediction model generated by using as input values an average value of time-series data of angles of ankle joints of one leg during 84% to 89% of one walking cycle and using as output values whether or not a subject has a risk of falling.

The walking parameter detecting unit 112 detects time-series data of the angle of the ankle joint of one leg during 84% to 89% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of angles of ankle joints of one leg in a period of 84% to 89% of one walking cycle.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using an average value of time-series data of angles of ankle joints of one leg during 84% to 89% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the angle of the ankle joint of one leg during 84% to 89% of one walking cycle to the prediction model, and thereby obtains a determination result indicating whether or not the test subject has a fall risk from the prediction model.

Fig. 19 is a diagram showing, in a sixth modification of the present embodiment, the average of the average values of the time-series data of the angles of the ankle joint of one leg of the subjects who do not have a risk of falling during the period of 84% to 89% of one walking cycle and the average value of the time-series data of the angles of the ankle joint of one leg of the subjects who do not have a risk of falling during the period of 84% to 89% of one walking cycle.

As shown in fig. 19, the average of the time-series data of the angle of the ankle joint of one leg of the subjects who had no risk of falling during 84% to 89% of one walking cycle was 13.7 degrees on average, and the average of the time-series data of the angle of the ankle joint of one leg of the subjects who had a risk of falling during 84% to 89% of one walking cycle was 11.4 degrees on average.

As described above, the average of the time-series data of the angle of the ankle joint of one leg of the subjects with a risk of falling is smaller than the average of the time-series data of the angle of the ankle joint of one leg of the subjects without a risk of falling during 84% to 89% of one walking cycle. For this purpose, a value between the average of the time-series data of the angle of the ankle joint of one leg of the subjects with a risk of falling during 84% to 89% of one walking cycle and the average of the time-series data of the angle of the ankle joint of one leg of the subjects without a risk of falling during 84% to 89% of one walking cycle, which are obtained through experiments, may be stored in the memory 12 as the threshold value. The fall risk determination unit 113 may determine the presence or absence of a fall risk by comparing an average value of time-series data of angles of one ankle joint of the test subject during 84% to 89% of one walking cycle with a threshold value stored in advance.

Next, a walking parameter of a seventh modification of the present embodiment will be described.

The walking parameter in the seventh modification of the present embodiment may be an average value of time-series data of a displacement of the waist in the vertical direction during the first period of the stance phase of one leg and an average value of time-series data of an angle of the knee joint during the second period of the stance phase of one leg.

In a seventh modification of the present embodiment, time-series data of displacements of the waist in the vertical direction of each of the plurality of test subjects and time-series data of the angle of one knee joint of each of the plurality of test subjects are detected, as in the above-described experiment. In the experiment, one standardized walking cycle was divided into ten sections, and the average value of the displacement of the waist in the vertical direction and the average value of the angle of one knee joint in one section or two or more continuous sections were calculated for each test subject. Then, a plurality of prediction models are created, each of which has, as an objective variable, whether or not the test subject is at risk of falling, and has, as explanatory variables, an average value of the displacement of the waist in the vertical direction in one or two or more consecutive intervals, and an average value of the angle of one knee joint. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve is calculated for each of the plurality of prediction models. Then, AUC values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In a seventh modification of the present embodiment, the AUC value of the prediction model, in which the average value of the displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and the average value of the angle of one knee joint during 1% to 60% of one walking cycle are taken as explanatory variables, is the highest.

The prediction model of the seventh modification of the present embodiment is created with the average of the average values of the displacements of the waist in the vertical direction during 1% to 40% of one walking cycle and the average value of the angles of the knee joint during 1% to 60% of one walking cycle as explanatory variables, with the presence or absence of the fall risk of the test subject being used as the objective variables. In fig. 20, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 20 is a curve that depicts the true positive rate and the false positive rate of a prediction model that has as explanatory variables the average value of the displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and the average value of the angle of one knee joint during 1% to 60% of one walking cycle. The AUC value of the ROC curve shown in fig. 20 is 0.734. In this case, the average value of the displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and the average value of the angle of one knee joint during 1% to 60% of one walking cycle are determined as walking parameters. Then, a prediction model created using as explanatory variables the average value of the displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and the average value of the angle of one knee joint during 1% to 60% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detection unit 112 detects time-series data of a displacement of the waist in the vertical direction during a first period of the stance phase of one leg and time-series data of an angle of the knee joint during a second period of the stance phase of one leg. The first period is a period of 1% to 40% of one walking cycle, and the second period is a period of 1% to 60% of one walking cycle. The walking parameter detecting unit 112 detects time-series data of a displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and time-series data of an angle of one knee joint during 1% to 60% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of vertical displacement of the waist during 1% to 40% of one walking cycle and an average value of time-series data of angle during 1% to 60% of one knee joint during one walking cycle.

The memory 12 stores in advance a prediction model generated by using as input values an average value of time-series data of displacement of the waist in the vertical direction during a first period of the stance phase of one leg and an average value of time-series data of angles of the knee joints during a second period of the stance phase of one leg, and using as output values whether or not the test subject is at risk of falling. The memory 12 stores in advance a prediction model generated by using as input values an average value of displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and an average value of angles of one knee joint during 1% to 60% of one walking cycle, and using as output values whether or not the test subject has a risk of falling.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using the average value of the time-series data of the displacement of the waist in the vertical direction and the average value of the time-series data of the angle of the knee joint.

The fall risk determination unit 113 determines whether the subject is at risk of falling using an average value of time-series data of the displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and an average value of time-series data of the angle of one knee joint during 1% to 60% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the displacement of the waist in the vertical direction during 1% to 40% of one walking cycle and the average value of the time-series data of the angle during 1% to 60% of one knee joint during one walking cycle into the prediction model, and thereby obtains a determination result indicating whether the subject has a fall risk from the prediction model.

As described above, the AUC value of the prediction model created solely using the displacement of the waist in the vertical direction at the stance phase was 0.733, and the AUC value of the prediction model created solely using the angle of the knee joint at the stance phase was 0.542. In contrast, the AUC value of the prediction model created using the displacement of the waist in the vertical direction during the stance phase and the angle of the knee joint during the stance phase was 0.734. Therefore, the prediction model created using the displacement of the waist in the vertical direction during the stance phase and the angle of the knee joint during the stance phase can determine the presence or absence of the risk of falling with high accuracy, as compared with a prediction model created using the displacement of the waist in the vertical direction during the stance phase and the angle of the knee joint during the stance phase, respectively.

Next, walking parameters of an eighth modification of the present embodiment will be described.

The walking parameter in the eighth modification of the present embodiment may be an average value of time-series data of displacement of the waist in the vertical direction during 9% to 19% of one walking cycle of one leg and an angle of the knee joint at 35% of the time of one walking cycle of one leg.

In the eighth modification of the present embodiment, as in the above-described experiment, time-series data of displacements of the waist in the vertical direction of each of the plurality of test subjects and time-series data of the angle of one knee joint of each of the plurality of test subjects are detected. Then, a prediction model is created in which whether the test subject has a fall risk or not is used as an objective variable, and an average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of one knee joint at the time 35% of one walking cycle are used as explanatory variables. The predictive model was evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve of the prediction model is calculated. Then, AUC values of ROC curves of the prediction models are calculated.

The prediction model of the eighth modification of the present embodiment is created with the target variables of whether or not the test subject is at risk of falling, and the explanatory variables of the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of one knee joint at 35% of the time of one walking cycle. In fig. 21, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 21 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created with the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of one knee joint at the time of 35% of one walking cycle as explanatory variables. The AUC value of the ROC curve shown in fig. 21 is 0.8109. In this case, the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of one knee joint at the time of 35% of one walking cycle are determined as walking parameters. A prediction model created using as explanatory variables the average value of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of one knee joint at the time 35% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detection unit 112 detects time-series data of a displacement of the waist in the vertical direction during a first period of the stance phase of one leg and time-series data of an angle of the knee joint during a second period of the stance phase of one leg. Specifically, the walking parameter detecting unit 112 detects time-series data of displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and an angle of 35% of one knee joint during one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of the vertical displacement of the waist in a period of 9% to 19% of one walking cycle.

The memory 12 stores in advance a prediction model generated by using as input values an average value of displacements of the waist in the vertical direction during 9% to 19% of one walking cycle and an angle of one knee joint at a time 35% of one walking cycle, and using as output values whether or not the test subject has a risk of falling.

The fall risk determination unit 113 determines whether the subject has a risk of falling, using the average value of the time-series data of the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle at the time 35% of one knee joint in one walking cycle. The fall risk determination unit 113 inputs, to the prediction model, an average value of time-series data of displacements of the waist in the vertical direction during 9% to 19% of one walking cycle and an angle of 35% of the time of one knee joint in one walking cycle, and thereby obtains a determination result indicating whether or not the test subject is at risk of falling from the prediction model.

As described above, the AUC value of the prediction model created solely using the displacement of the waist in the vertical direction during 9% to 19% of one gait cycle was 0.8058, and the AUC value of the prediction model created solely using the angle of the knee joint at the time of 35% of one gait cycle was 0.6242. In contrast, the AUC value of the prediction model created using the displacement of the waist in the vertical direction during 9% to 19% of one gait cycle and the angle of the knee joint at the time of 35% of one gait cycle was 0.8109. Therefore, the prediction model created using the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of the knee joint at 35% of one walking cycle can determine the presence or absence of the risk of falling with high accuracy, as compared to the prediction model created using the displacement of the waist in the vertical direction during 9% to 19% of one walking cycle and the angle of the knee joint at 35% of one walking cycle, respectively.

Next, walking parameters of a ninth modification of the present embodiment will be described.

The walking parameter in the ninth modification of the present embodiment may be an average value of time-series data of a vertical displacement of the waist during the first period of the stance phase of one leg, an average value of time-series data of a vertical displacement of the waist during the second period of the stance phase of one leg, or an average value of time-series data of an angle of the ankle joint during the third period of the stance phase of one leg.

In the ninth modification of the present embodiment, as in the above-described experiment, time-series data of vertical displacements of the waist of each of the plurality of test subjects and time-series data of the angle of the ankle joint of each of the plurality of test subjects are detected. In the experiment, one standardized walking cycle was divided into ten sections, and the average value of the displacement of the waist in the vertical direction in one section or two or more continuous sections and the average value of the angle of one ankle joint in one section or two or more continuous sections were calculated for each test subject. Then, a plurality of prediction models are created, which have, as objective variables, whether the test subject is at risk of falling, and use, as explanatory variables, the average of the vertical displacement of the waist in one section or two or more consecutive sections and the average of the angle of one ankle joint in one section or two or more consecutive sections. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve is calculated for each of the plurality of prediction models. Then, AUC values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In a ninth modification of the present embodiment, the AUC value of the prediction model having the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle as explanatory variables is the highest.

The prediction model of the ninth modification of the present embodiment is created with the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle as the target variables. In fig. 22, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 22 is a curve that depicts the true positive rate and the false positive rate of a prediction model created with the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle as explanatory variables. The AUC value of the ROC curve shown in fig. 22 was 0.746. In this case, the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle are determined as walking parameters. In addition, a prediction model created using as explanatory variables the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detection unit 112 detects time-series data of a vertical displacement of the waist during a first period of the stance phase of one leg, time-series data of a vertical displacement of the waist during a second period of the foot suspension phase of one leg, and time-series data of an angle of the ankle joint during a third period of the foot suspension phase of one leg. The first period is a period of 1% to 60% of one walking cycle, the second period is a period of 61% to 80% of one walking cycle, and the third period is a period of 61% to 100% of one walking cycle. The walking parameter detecting unit 112 detects time-series data of displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, time-series data of displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and time-series data of an angle of one ankle joint during 61% to 100% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of vertical displacement of the waist during 1% to 60% of one walking cycle, an average value of time-series data of vertical displacement of the waist during 61% to 80% of one walking cycle, and an average value of time-series data of angle of one ankle joint during 61% to 100% of one walking cycle.

The memory 12 stores in advance a prediction model generated by using as input values an average value of time-series data of displacements of the waist in the vertical direction during a first period of the stance period of one leg, an average value of time-series data of displacements of the waist in the vertical direction during a second period of the stance period of one leg, and an average value of time-series data of angles of ankle joints during a third period of the stance period of one leg, and using as output values whether or not the test subject is at risk of falling. The memory 12 stores in advance a prediction model generated by using as input values an average value of displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, an average value of displacement of the waist in the vertical direction during 61% to 80% of one walking cycle, and an average value of angles of one ankle joint during 61% to 100% of one walking cycle, and using as output values whether or not the subject has a risk of falling.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using the average value of the time-series data of the vertical displacement of the waist and the average value of the time-series data of the angle of the ankle joint.

The fall risk determination unit 113 determines whether the test subject has a risk of falling, using an average value of time-series data of vertical displacement of the waist during 1% to 60% of one walking cycle, an average value of time-series data of vertical displacement of the waist during 61% to 80% of one walking cycle, and an average value of time-series data of an angle of one ankle during 61% to 100% of one walking cycle. The fall risk determination unit 113 obtains a determination result indicating whether the test subject is at risk of falling from the prediction model by inputting, to the prediction model, an average value of time-series data of displacements of the waist in the vertical direction during 1% to 60% of one walking cycle, an average value of time-series data of displacements of the waist in the vertical direction during 61% to 80% of one walking cycle, and an average value of time-series data of angles of one ankle joint during 61% to 100% of one walking cycle.

As described above, the AUC value of the prediction model created solely using the displacement of the waist in the vertical direction in the stance phase was 0.733, and the AUC value of the prediction model created solely using the angle of the ankle joint in the foot suspension phase was 0.595. In contrast, the AUC value of the prediction model created using the displacement of the waist in the vertical direction during the first period of the stance phase, the displacement of the waist in the vertical direction during the second period of the foot suspension phase, and the angle of the ankle joint during the third period of the foot suspension phase was 0.746. Therefore, the prediction model created using the displacement of the waist in the vertical direction during the stance phase, the displacement of the waist in the vertical direction during the second period of the stance phase, and the angle of the ankle joint during the third period of the stance phase can determine the presence or absence of the risk of falling with high accuracy, as compared with the prediction model created using the displacement of the waist in the vertical direction during the stance phase and the angle of the ankle joint during the foot suspension phase, respectively.

Next, walking parameters of a tenth modification of the present embodiment will be described.

The walking parameter in the tenth modification of the present embodiment may be an average value of time-series data of an angle of the knee joint in the first period of the foot standing period of one leg and an average value of time-series data of an angle of the ankle joint in the second period of the foot floating period of one leg.

In a tenth modification of the present embodiment, time-series data of the angle of one knee joint of each of a plurality of test subjects and time-series data of the angle of one ankle joint of each of a plurality of test subjects are detected, as in the above-described experiment. In the experiment, one standardized walking cycle was divided into ten sections, and the average value of the angles of one knee joint in one section or two or more continuous sections and the average value of the angles of one ankle joint in one section or two or more continuous sections were calculated for each test subject. Then, a plurality of prediction models are created, which take whether the test subject has a fall risk as a target variable, and take the average value of the angles of one knee joint in one section or two or more continuous sections and the average value of the angles of one ankle joint in one section or two or more continuous sections as explanatory variables. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve is calculated for each of the plurality of prediction models. Then, AUC values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In a tenth modification of the present embodiment, the AUC value of the prediction model in which the average value of the angles of one knee joint during 1% to 60% of one walking cycle and the average value of the angles of one ankle joint during 61% to 100% of one walking cycle are used as explanatory variables is the highest.

The prediction model according to the tenth modification of the present embodiment is created with the average value of the angles of one knee joint during 1% to 60% of one walking cycle and the average value of the angles of one ankle joint during 61% to 100% of one walking cycle as explanatory variables, with the target variables being whether the subject is at risk of falling. In fig. 23, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 23 is a curve that depicts the true positive rate and the false positive rate of a prediction model that is created using as explanatory variables the average value of the angles of one knee joint during 1% to 60% of one walking cycle and the average value of the angles of one ankle joint during 61% to 100% of one walking cycle. The AUC value of the ROC curve shown in fig. 23 is 0.628. In this case, the average value of the angle of one knee joint during 1% to 60% of one walking cycle and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle are determined as walking parameters. Furthermore, a prediction model created using as explanatory variables the average value of the angles of one knee joint during 1% to 60% of one walking cycle and the average value of the angles of one ankle joint during 61% to 100% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detection unit 112 detects time-series data of an angle of the knee joint in a first period of the foot standing period of one leg and time-series data of an angle of the ankle joint in a second period of the foot floating period of one leg. The first period is a period of 1% to 60% of one walking cycle, and the second period is a period of 61% to 100% of one walking cycle. The walking parameter detecting unit 112 detects time-series data of an angle of one knee joint during 1% to 60% of one walking cycle and time-series data of an angle of one ankle joint during 61% to 100% of one walking cycle. The walking parameter detecting unit 112 calculates an average value of time-series data of angles of one knee joint in a period of 1% to 60% of one walking cycle and an average value of time-series data of angles of one ankle joint in a period of 61% to 100% of one walking cycle.

The memory 12 stores in advance a prediction model generated by using as input values an average value of time-series data of angles of a knee joint in a first period of a foot standing period of one leg and an average value of time-series data of angles of an ankle joint in a second period of a foot floating period of one leg, and using as output values whether or not a subject has a risk of falling. The memory 12 stores in advance a prediction model generated by using as input values an average value of angles of one knee joint during 1% to 60% of one walking cycle and an average value of angles of one ankle joint during 61% to 100% of one walking cycle, and using as output values whether or not a subject has a risk of falling.

The fall risk determination unit 113 determines whether the subject is at risk of falling, using the average value of the time-series data of the angle of the knee joint and the average value of the time-series data of the angle of the ankle joint.

The fall risk determination unit 113 determines whether the subject is at risk of falling using an average value of time-series data of angles of one knee joint in a period of 1% to 60% of one walking cycle and an average value of time-series data of angles of one ankle joint in a period of 61% to 100% of one walking cycle. The fall risk determination unit 113 inputs the average value of the time-series data of the angle of one knee joint in a period of 1% to 60% of one walking cycle and the average value of the time-series data of the angle of one ankle joint in a period of 61% to 100% of one walking cycle to the prediction model, and thereby obtains a determination result indicating whether or not the subject is at risk of falling from the prediction model.

As described above, the AUC values of the prediction models created using the angle of the knee joint and the angle of the ankle joint individually were 0.542 and 0.595, and the AUC value of the prediction model created using the angle of the knee joint and the angle of the ankle joint was 0.628. Therefore, the prediction model created using the angle of the knee joint and the angle of the ankle joint can determine the presence or absence of the risk of falling with high accuracy, as compared with the prediction model created using the angle of the knee joint and the angle of the ankle joint individually.

Next, walking parameters of an eleventh modification of the present embodiment will be described.

The walking parameter in the eleventh modification of the present embodiment may be an average value of time-series data of a displacement of the waist in the vertical direction during the first period of the stance phase of one leg, an average value of time-series data of an angle of the knee joint during the second period of the stance phase of one leg, or an average value of time-series data of an angle of the ankle joint during the third period of the foot suspension phase of one leg.

In the eleventh modification of the present embodiment, as in the above-described experiment, time-series data of the vertical displacement of the waist of each of the plurality of test subjects, time-series data of the angle of one knee joint of each of the plurality of test subjects, and time-series data of the angle of the ankle joint of each of the plurality of test subjects are detected. In the experiment, one standardized walking cycle was divided into ten sections, and the average value of the displacement in the vertical direction of the waist in one section or two or more continuous sections, the average value of the angle of one knee joint in one section or two or more continuous sections, and the average value of the angle of one ankle joint in one section or two or more continuous sections were calculated for each test subject. Then, a plurality of prediction models are created, which have, as objective variables, whether the test subject is at risk of falling or not, and as explanatory variables, the average value of the displacement in the vertical direction of the waist in one section or two or more consecutive sections, the average value of the angle of one knee joint in one section or two or more consecutive sections, and the average value of the angle of one ankle joint in one section or two or more consecutive sections. Multiple predictive models are evaluated by cross-validation. As the cross-validation, leave-one-out cross-validation is adopted. Then, an ROC curve is calculated for each of the plurality of prediction models. Then, AUC values of ROC curves of the plurality of prediction models are calculated, and the prediction model having the highest AUC value is selected.

In the eleventh modification of the present embodiment, the AUC value of the prediction model having the explanatory variables of the average value of the vertical displacement of the waist during 1% to 60% of one walking cycle, the average value of the angle of one knee joint during 1% to 60% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle is the highest.

The prediction model of the eleventh modification of the present embodiment is created with the average value of the vertical displacement of the waist during 1% to 60% of one walking cycle, the average value of the angle of one knee joint during 1% to 60% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle as explanatory variables, and with the presence or absence of a fall risk of the test subject as objective variables. In fig. 24, the vertical axis represents the true positive rate, and the horizontal axis represents the false positive rate. The true positive rate indicates a proportion of test subjects having a risk of falling that are correctly determined by the prediction model as having a risk of falling, and the false positive rate indicates a proportion of test subjects having no risk of falling that are incorrectly determined by the prediction model as having a risk of falling.

The ROC curve shown in fig. 24 is a curve that depicts the true positive rate and the false positive rate of a prediction model created using as explanatory variables the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the angle of one knee joint during 1% to 60% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle. The AUC value of the ROC curve shown in fig. 24 was 0.691. In this case, the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the angle of one knee joint during 1% to 60% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle are determined as the walking parameters. A prediction model created using as explanatory variables the average value of the displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, the average value of the angle of one knee joint during 1% to 60% of one walking cycle, and the average value of the angle of one ankle joint during 61% to 100% of one walking cycle is determined as the prediction model used by the fall risk determination unit 113.

The walking parameter detection unit 112 detects time-series data of a vertical displacement of the waist in a first period of the stance phase of one leg, time-series data of an angle of the knee joint in a second period of the stance phase of one leg, and time-series data of an angle of the ankle joint in a third period of the foot suspension phase of one leg. The first period is a period of 1% to 60% of one walking cycle, the second period is a period of 1% to 60% of one walking cycle, and the third period is a period of 61% to 100% of one walking cycle. The walking parameter detecting unit 112 detects time-series data of a displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, time-series data of an angle of one knee joint during 1% to 60% of one walking cycle, and time-series data of an angle of one ankle joint during 61% to 100% of one walking cycle.

The walking parameter detecting unit 112 calculates an average value of time-series data of vertical displacement of the waist during 1% to 60% of one walking cycle, an average value of time-series data of an angle of one knee joint during 1% to 60% of one walking cycle, and an average value of time-series data of an angle of one ankle joint during 61% to 100% of one walking cycle.

The memory 12 stores in advance a prediction model generated by using as input values an average value of time-series data of a displacement of the waist in the vertical direction during a first period of the stance phase of one leg, an average value of time-series data of an angle of the knee joint during a second period of the stance phase of one leg, and an average value of time-series data of an angle of the ankle joint during a third period of the foot suspension phase of one leg, and using as output values whether or not the test subject is at risk of falling. The memory 12 stores in advance a prediction model generated by using as input values an average value of displacement of the waist in the vertical direction during 1% to 60% of one walking cycle, an average value of angles of one knee joint during 1% to 60% of one walking cycle, and an average value of angles of one ankle joint during 61% to 100% of one walking cycle, and using as output values whether or not the subject is at risk of falling.

The fall risk determination unit 113 determines whether the test subject is at risk of falling, using the average value of the time-series data of the displacement of the waist in the vertical direction, the average value of the time-series data of the angle of the knee joint, and the average value of the time-series data of the angle of the ankle joint.

The fall risk determination unit 113 determines whether the subject is at risk of falling using an average value of time-series data of vertical displacement of the waist during 1% to 60% of one walking cycle, an average value of time-series data of angles during 1% to 60% of one walking cycle of one knee joint, and an average value of time-series data of angles during 61% to 100% of one walking cycle of one ankle joint. The fall risk determination unit 113 obtains a determination result indicating whether the test subject is at risk of falling from the prediction model by inputting, to the prediction model, an average value of time-series data of displacements of the waist in the vertical direction during 1% to 60% of one walking cycle, an average value of time-series data of angles of one knee joint during 1% to 60% of one walking cycle, and an average value of time-series data of angles of one ankle joint during 61% to 100% of one walking cycle.

As described above, the AUC values of the prediction models created using the angle of the knee joint and the angle of the ankle joint individually were 0.542 and 0.595, and the AUC value of the prediction model created using the displacement of the waist in the vertical direction, the angle of the ankle joint, and the angle of the knee joint was 0.691. Therefore, the prediction model created using the displacement of the waist in the vertical direction, the angle of the knee joint, and the angle of the ankle joint can determine the presence or absence of the risk of falling with high accuracy, as compared with the prediction model created using the angle of the knee joint and the angle of the ankle joint individually.

Fig. 25 is a schematic diagram showing an example of an evaluation result screen displayed in the present embodiment.

The display unit 3 displays an evaluation result screen shown in fig. 25. The evaluation result screen includes a fall risk evaluation presentation area 31 and an evaluation message 32 that indicate the evaluation value of the past fall risk and the evaluation value of the current fall risk. In the fall risk assessment presentation area 31 of fig. 25, it is displayed that the assessment of the fall risk is performed once a month, the assessment values of the fall risk for the past six months, and the assessment values of the fall risk for the present month.

The evaluation value of fall risk is a value representing the degree of fall risk calculated by a predictive model. The value representing the degree of fall risk is represented by, for example, 0.0 to 1.0. The evaluation result presentation unit 114 presents a value indicating the degree of fall risk converted into a percentage as an evaluation value of fall risk.

In addition, in the case where the evaluation value of the present fall risk is displayed together with the evaluation value of the past fall risk, the fall risk determination section 113 stores the evaluation value of the fall risk in the memory 12.

Also, the fall risk assessment presentation area 31 may also display whether or not the test subject has a fall risk as an assessment result.

Also, an evaluation message 32 can be displayed that "the fall risk is reduced from the last month, a good state is maintained, please keep living in this way". The evaluation result presentation section 114, in a case where the evaluation value of the fall risk in this month is lower than the evaluation value of the fall risk in the previous month, and the evaluation value of the fall risk in this month is lower than 0.5, reads out the evaluation message 32 shown in fig. 25 from the memory 12 and outputs it to the display section 3.

In the present embodiment, the evaluation value of the current fall risk and the evaluation value of the past fall risk are displayed, but the present invention is not limited to this, and only the evaluation value of the current fall risk may be displayed. In this case, the fall risk determining section 113 does not need to store the evaluation value of the fall risk in the memory 12.

The camera 2 of the present embodiment may be a security camera installed in front of the entrance, a slave camera of an intercom, or a monitoring camera installed indoors. The display unit 3 may be a monitor of a smartphone, a tablet computer, or an intercom.

In the present embodiment, the walking parameter detection unit 112 extracts the bone data based on the moving image data acquired from the camera 2, but the present invention is not limited to this, and the bone data may be extracted by a motion capture system (motion capture system). The motion capture system may operate based on any of optical, magnetic, mechanical, and inertial sensor types. For example, an optical motion capture system captures an image of a test object with a label attached to a joint portion with a camera, and detects the position of the label from the captured image. The walking parameter detection unit 112 acquires skeletal data of the test subject from the position data detected by the motion capture system. As the optical motion capture system, for example, a three-dimensional motion analysis device manufactured by intet Reha co, Ltd.

The motion capture system may further include a depth sensor and a color camera, and may automatically extract position information of an articulation point of the test object from the video image and detect the posture of the test object. In this case, the test object does not need to be labeled. As such a motion capture system, for example, Kinect manufactured by microsoft corporation can be used.

When the walking motion is measured by the motion capture system, it is preferable to extract the angle of the ankle joint, the angle of the knee joint, or the vertical displacement of the waist during the walking motion from the position coordinates, and detect the feature amount of the walking motion from the extracted angle or displacement.

In the above embodiments, each component may be configured by dedicated hardware, or may be realized by executing a software program suitable for each component. Each component may be realized by causing a program execution unit such as a CPU or a processor to read a software program recorded in a recording medium such as a hard disk or a semiconductor memory.

A part or all of the functions of the apparatus according to the embodiment of the present invention may be typically realized as an lsi (large Scale integration) integrated circuit. Some or all of these functions may be formed into chips, or may be formed into chips including some or all of these functions. The integrated circuit is not limited to the LSI, and may be realized by a dedicated circuit or a general-purpose processor. An fpga (field Programmable Gate array) which can be programmed after LSI manufacturing or a reconfigurable processor which can reconfigure connection or setting of circuit cells inside LSI can be used.

In addition, a part or all of the functions of the apparatus according to the embodiment of the present invention may be realized by causing a processor such as a CPU to execute a program.

Further, the numbers used in the above are examples given for specifically explaining the present invention, and the present invention is not limited to these exemplified numbers.

The order in which the steps shown in the flowcharts are executed is merely an example given for specifically explaining the present invention, and may be an order other than the above, as long as the same effects can be obtained. Moreover, some of the above steps may be performed concurrently with (or in parallel with) other steps.

The technique according to the present invention is useful as a technique for evaluating a fall risk based on a walking motion of a test subject because the fall risk can be evaluated easily and with high accuracy.

53页详细技术资料下载

上一篇：一种医用注射器针头装配设备

下一篇：头部转动角度测量装置

Fall risk assessment method, fall risk assessment device, and recording medium

相关技术

网友询问留言