阅读说明：本技术 一种入睡潜伏期细化分类系统 (Sleep latency refining and classifying system ) 是由 李建军 段韩路 于 2019-09-24 设计创作，主要内容包括：本发明涉及一种入睡潜伏期细化分类系统。现有入睡潜伏期细化分类系统计算过程复杂且分类精度较低。本发明包括处理器和基础数据采集器,处理器通过基础数据采集器采集用户的基础数据,并通过计算获得用户处于入睡潜伏期时的细分睡眠状态,既能为后续助眠设备提供运行依据,还简化系统结构,提升计算效率和运行能耗,又提升穿戴舒适性,提升使用体验。(The invention relates to a sleep latency refining and classifying system. The existing sleep-in latent period refining and classifying system is complex in calculation process and low in classifying precision. The sleep aid system comprises the processor and the basic data collector, wherein the processor collects basic data of a user through the basic data collector and obtains the subdivided sleep state of the user in the sleep aid latency period through calculation, so that the operation basis can be provided for subsequent sleep aid equipment, the system structure is simplified, the calculation efficiency and the operation energy consumption are improved, the wearing comfort is improved, and the use experience is improved.)

1. A sleep-in latent period refining and classifying system is characterized in that a processor collects basic data of a user through a basic data collector and obtains a sleep state subdivision state of the user in a sleep-in latent period through calculation, and the sleep state refining and classifying system comprises the following steps:

the method comprises the steps that firstly, basic data are obtained through a basic data collector, in the collection process, the basic data are obtained through an interval collection mode by the basic data collector, and single-frame data packets are formed by the basic data which are continuously collected at a single time;

secondly, the processor carries out filtering processing on the single-frame data packet and sequentially obtains an Alpha wave oscillogram, a Beta wave oscillogram, a Theta wave oscillogram and an SEEG wave oscillogram with frequencies of 8-12 Hz, 18-30 Hz, 4-7 Hz and 40-50 Hz respectively;

thirdly, counting the numerical values of the acquisition points in each oscillogram, and sequentially carrying out absolute value processing and data smoothing processing to obtain a primary characteristic value parameter group and an auxiliary characteristic value parameter group through accumulation calculation, wherein the primary characteristic value parameter group comprises an Alpha value, a Beta value, a Theta value and a SEEG value, and the auxiliary characteristic value parameter group comprises a Num-ARI value, a Num-LCZ value, a Num-Alpha value, a Num-Beta value, a Num-Theta value and a Num-EOG value;

fourthly, calculating to obtain a secondary characteristic value parameter group through the primary characteristic value parameter group, wherein the secondary characteristic value parameter group comprises an AVB value, an AVS value, a BVS value, a TVA value, a TVB value and a TVS value;

and fifthly, finely classifying the sleep state of the user in the sleep-in latent period through the auxiliary characteristic value parameter group and the secondary characteristic value parameter obtained in the third step and the fourth step, so that the sleep state of the user is classified into a waking active state, an attention transfer state or a waking quiet state.

2. The system for refining and classifying sleep-onset latencies according to claim 1, wherein the processor classifies the sleep states of the user through a decision tree, wherein at least six nodes are set on the decision tree for classifying the sleep states, and specifically:

setting a Threshold Num-ARI-Threshold at a first node, when Num-ARI is larger than Num-ARI-Threshold, judging the single-frame data packet as an artificial artifact, and outputting a sleep state obtained when the previous single-frame data packet is processed to the outside through a processor, otherwise, switching to a second node;

setting a Threshold value AVS-Threshold at a second node, when the AVS is larger than the AVS-Threshold, judging that the user is in a sleep latency period or a rapid eye movement period by a processor, and switching to a fifth node, otherwise, judging that the user is in a light sleep period, a light sleep period or a deep sleep period by the processor, and switching to a third node;

at a third node, setting Threshold values TVS-Threshold and Num-Theta-Threshold, when TVS is larger than TVS-Threshold and Num-Theta is larger than Num-Theta-Threshold, judging that the user is in a light sleep period by a processor, switching to a fourth node, otherwise, judging that the user is in a light sleep period or a deep sleep period by the processor, stopping the system, and recording the single-frame data packet as a data packet of the falling-asleep frame;

at a fourth node, when the processor classifies at least 3 continuous single-frame data packets as a light sleep period, the system stops running, and records the first single-frame data packet classified as the light sleep period as a sleep frame data packet, otherwise, the first node is switched to and processes the next single-frame data packet;

setting Threshold values TVA-Threshold and Num-EOG-Threshold at a fifth node, when TVA > TVA-Threshold and Num-EOG > Num-EOG-Threshold, judging that the user is in a fast eye movement period by a processor, stopping the system, recording the single frame data packet as a sleep-in frame data packet, and otherwise, judging that the user is in a sleep-in latent period by the processor and transferring to a sixth node;

setting Threshold values AVB-Threshold, Num-Alpha-Threshold and Num-Beta-Threshold at the sixth node,

when AVB < AVB-Threshold and Num-Beta > Num-Beta-Threshold, the processor determines that the user is in an awake active state,

when AVB > AVB-Threshold and Num-Alpha > Num-Alpha-Threshold, the processor determines that the user is in an awake and quiet state,

otherwise, the processor determines that the user is in a state of attention diversion.

3. The system for refining and classifying sleep-onset latencies according to claim 2, wherein at the first node, a Threshold Num-LCZ-Threshold is set, when Num-LCZ > Num-LCZ-Threshold, the single-frame packet is determined as an artificial artifact and a sleep state obtained when a previous single-frame packet is processed is outputted to the outside through the processor, otherwise, the second node is shifted to; or setting a Threshold BVS-Threshold at the second node, and when the BVS is more than the BVS-Threshold, the processor judges that the user is in a sleep latency period or a rapid eye movement period and shifts to the fifth node, otherwise, the processor judges that the user is in a light sleep period, a light sleep period or a deep sleep period and shifts to the third node; or setting Threshold values TVB-Threshold and Num-EOG-Threshold, when TVB > TVA-Threshold and Num-EOG > Num-EOG-Threshold, judging that the user is in a rapid eye movement period by the processor, stopping the system, recording the single frame data packet as a sleep frame data packet, otherwise, judging that the user is in a sleep latency period by the processor, and switching to the sixth node.

4. The system for refining and classifying sleep-onset latencies according to claim 2 or 3, wherein the collection time interval between adjacent single-frame data packets is A, A is more than or equal to 30s and less than or equal to 60s, the single continuous collection time required for forming a single-frame data packet is B, B is more than or equal to A, and B is more than or equal to 5s and less than or equal to 60s, the data collection frequency of the basic data collector is C, C is more than or equal to 200samples/s and less than or equal to 1000samples/s, and the data collection times of the basic data collector in forming a single-frame data packet are n and n is C.

5. The system for refining classification of sleep latency according to claim 4, wherein the primary feature value parameter group is obtained by the following steps:

firstly, respectively carrying out absolute value processing on numerical values corresponding to each acquisition point in Alpha wave oscillogram, Beta wave oscillogram, Theta wave oscillogram and SEEG wave oscillogram in sequence, and obtaining a data set X_m；

Thereafter, data set X is processed_mCarrying out data smoothing processing and obtaining a transition data set Y_m，The data length of the transition data group is m, m is more than or equal to 1 and less than or equal to n-0.5C, wherein 0.5C is the length of data smoothing processing;

finally, for the transition data set Y_mAnd performing accumulation calculation and sequentially obtaining the Alpha value, the Beta value, the Theta value and the SEEG value.

6. The system of claim 5, wherein each parameter in the primary parameter set is divided by m to obtain a Mean-Alpha value, a Mean-Beta value, a Mean-Theta value, and a Mean-SEEG value, and then calculated to obtain an AVB value, an AVS value, a BVS value, a TVA value, a TVB value, and a TVS value,

AVB＝Mean-Alpha/Mean-Beta，

AVS＝Mean-Alpha/Mean-SEEG，

BVS＝Mean-Beta/Mean-SEEG，

TVA＝Mean-Theta/Mean-Alpha，

TVB＝Mean-Theta/Mean-Beta，

TVS＝Mean-Theta/Mean-SEEG。

7. the system for refining and classifying the sleep-onset latency according to claim 5, wherein a corresponding maximum threshold and a corresponding minimum threshold are set for each oscillogram, a data fluctuation range is formed, the number of acquisition points of each oscillogram, which exceed the data fluctuation range, is accumulated and counted, and a Num-ARI value is formed; or accumulating and counting the times of the curve in each oscillogram crossing the x axis to form a Num-LCZ value; or counting the occurrence time of Alpha waves in the Alpha wave oscillogram so as to obtain a Num-Alpha value; or counting the occurrence time of the Beta wave in the Beta wave oscillogram so as to obtain a Num-Beta value; or counting the occurrence time of the Theta wave in the Theta wave oscillogram so as to obtain the Num-Theta value.

8. The system for refining and classifying sleep-onset latency according to claim 5, wherein an eye movement threshold is set, an eye electrical pulse signal is obtained from a SEEG wave waveform diagram, the number of times that the amplitude of the eye electrical pulse signal exceeds the eye movement threshold is counted to obtain a Num-EOG value, and the eye movement pulse signal is obtained by the following steps:

firstly, counting the values of each acquisition point of the SEEG wave oscillogram to form a basic value group with n number;

sequentially extracting X window value groups from the basic value group, wherein the window value groups comprise L basic values sequentially intercepted along the basic value group, X is n-L +1, L is less than n, and L is an odd number;

then, L basic values in each window value group are sequenced from small to large, and the (L +1)/2 th basic value is set as a smooth value of the window value group;

then, forming a smooth value group by the smooth values obtained in each window value group, wherein the smooth value group comprises X smooth values;

and finally, subtracting the numerical values of the corresponding sequence in the smooth numerical value set from the X numerical values in the middle of the basic numerical value set to form X difference values, wherein the difference values form the eye electrical pulse signals.

9. The sleep latency refinement classification system as claimed in claim 1, comprising,

the basic data collector is used for obtaining basic data through the portable wearable equipment and is provided with an electric signal collecting assembly;

and the processor receives the data from the basic data acquisition unit and calculates the sleep state subdivision when the user is in the sleep latency period.

10. The system for refining and classifying the sleep latency period as claimed in claim 9, wherein the basic data collector is a pair of glasses (1), the glasses (1) comprise a frame (2) and legs (3), the electrical signal collecting assembly comprises a bridge reference electrode (6) arranged in the middle of the frame (2) and a left front temporal electrode (4) and a right front temporal electrode (5) respectively arranged at two ends of the frame (2); or, the basic data collector is an eye patch (7), the eye patch (7) comprises a cover body and a binding band, and the electric signal collection assembly comprises a frontal pole reference electrode (10) arranged in the middle of the cover body, and a left frontal pole electrode (8) and a right frontal pole electrode (9) arranged at two ends of the cover body.

Technical Field

The invention relates to the field of sleep, in particular to a sleep latency refinement and classification system.

Background

The quality of sleep directly affects the health and even the life of people, and more people have sleep disorder problems because the rhythm of life is accelerated. Medical research shows that occasional insomnia can cause fatigue and uncoordinated movement on the next day, and long-term insomnia can cause the consequences of incapability of concentrating attention, disorder of memory, inattentive work and the like. Specifically, the insomnia is difficult to fall asleep, in order to improve the falling asleep quality and efficiency, electroencephalogram bioelectricity signals are collected through professional sleep-aid detection equipment, various characteristic parameters of the electroencephalogram bioelectricity signals are extracted through different analysis methods, finally, the sleep state of each sleep stage is identified through a classification algorithm, the sleep-aid quality is improved through a specific stimulation mode, and how to distinguish the states of each sleep stage in the falling sleep latency period becomes the key point for implementing the method. Polysomnography is currently an important new technique in sleep medicine, called the "gold standard" for diagnosing sleep disorders. However, since the multi-lead sleep monitor is used for monitoring, more sensors are needed, and most of the sensors are attached to the head and the trunk of a user, so that the monitoring environment is uncomfortable, the multi-lead sleep monitor is sensitive to adhesion, the sensors are easy to fall off, and the like, although the portable sleep monitor overcomes the defects of the multi-lead sleep monitor, such as less leads, convenience in carrying and the like, common consumers cannot conveniently use the monitor due to the factors of high power supply requirement, high price, accurate requirement on electrode positions, various operation steps and the like, and further popularization and application in the civil market are influenced.

Disclosure of Invention

In order to solve the defects of the prior art, the invention provides a refined classification system for the sleep-in latent period, a processor acquires data through a basic data acquisition unit and accurately distinguishes states of all stages of the sleep-in latent period, so that a running basis can be provided for subsequent sleep-assisting equipment, the system structure is simplified, the calculation efficiency and the running energy consumption are improved, the wearing comfort is improved, and the use experience is improved.

The invention is realized by the following modes: a processor acquires basic data of a user through a basic data acquisition unit and obtains a sleep state subdivision state of the user in a sleep latency period through calculation, wherein the sleep state subdivision system comprises the following steps:

the method comprises the steps that firstly, basic data are obtained through a basic data collector, in the collection process, the basic data are obtained through an interval collection mode by the basic data collector, and single-frame data packets are formed by the basic data which are continuously collected at a single time;

secondly, the processor carries out filtering processing on the single-frame data packet and sequentially obtains an Alpha wave oscillogram, a Beta wave oscillogram, a Theta wave oscillogram and an SEEG wave oscillogram with frequencies of 8-12 Hz, 18-30 Hz, 4-7 Hz and 40-50 Hz respectively;

thirdly, counting the numerical values of the acquisition points in each oscillogram, and sequentially carrying out absolute value processing and data smoothing processing to obtain a primary characteristic value parameter group and an auxiliary characteristic value parameter group through accumulation calculation, wherein the primary characteristic value parameter group comprises an Alpha value, a Beta value, a Theta value and a SEEG value, and the auxiliary characteristic value parameter group comprises a Num-ARI value, a Num-LCZ value, a Num-Alpha value, a Num-Beta value, a Num-Theta value and a Num-EOG value;

fourthly, calculating to obtain a secondary characteristic value parameter group through the primary characteristic value parameter group, wherein the secondary characteristic value parameter group comprises an AVB value, an AVS value, a BVS value, a TVA value, a TVB value and a TVS value;

and fifthly, finely classifying the sleep state of the user in the sleep-in latent period through the auxiliary characteristic value parameter group and the secondary characteristic value parameter obtained in the third step and the fourth step, so that the sleep state of the user is classified into a waking active state, an attention transfer state or a waking quiet state.

The basic data of limited types are collected by the basic data collector, a primary characteristic value parameter group, an auxiliary characteristic value parameter group and a secondary characteristic value parameter group are obtained through calculation of the processor on the basis of the basic data, and the subdivided sleep states of the users in the sleep-falling latent period are judged and classified, so that a reference basis is provided for subsequent sleep-helping operation. This system has the characteristics that use basic data kind few, calculation and classification process are simple, has both simplified basic data collection station's structure through reducing basic data collection kind, and convenience of customers dresses the use, promotes the data accuracy, still reduces the requirement to the treater through simplifying calculation and categorised process, promotes the arithmetic speed, reduces the hardware cost, ensures hardware operating stability, promotes and uses experience.

Preferably, the processor classifies the sleep state of the user through a decision tree, wherein at least six nodes are set on the decision tree to classify the sleep state, specifically:

setting a Threshold Num-ARI-Threshold at a first node, when Num-ARI is larger than Num-ARI-Threshold, judging the single-frame data packet as an artificial artifact, and outputting a sleep state obtained when the previous single-frame data packet is processed to the outside through a processor, otherwise, switching to a second node;

setting a Threshold value AVS-Threshold at a second node, when the AVS is larger than the AVS-Threshold, judging that the user is in a sleep latency period or a rapid eye movement period by a processor, and switching to a fifth node, otherwise, judging that the user is in a light sleep period, a light sleep period or a deep sleep period by the processor, and switching to a third node;

at a third node, setting Threshold values TVS-Threshold and Num-Theta-Threshold, when TVS is larger than TVS-Threshold and Num-Theta is larger than Num-Theta-Threshold, judging that the user is in a light sleep period by a processor, switching to a fourth node, otherwise, judging that the user is in a light sleep period or a deep sleep period by the processor, stopping the system, and recording the single-frame data packet as a data packet of the falling-asleep frame;

at a fourth node, when the processor classifies at least 3 continuous single-frame data packets as a light sleep period, the system stops running, and records the first single-frame data packet classified as the light sleep period as a sleep frame data packet, otherwise, the first node is switched to and processes the next single-frame data packet;

setting Threshold values TVA-Threshold and Num-EOG-Threshold at a fifth node, when TVA > TVA-Threshold and Num-EOG > Num-EOG-Threshold, judging that the user is in a fast eye movement period by a processor, stopping the system, recording the single frame data packet as a sleep-in frame data packet, and otherwise, judging that the user is in a sleep-in latent period by the processor and transferring to a sixth node;

setting Threshold values AVB-Threshold, Num-Alpha-Threshold and Num-Beta-Threshold at the sixth node,

the processor determines that the user is in an awake and active state when AVB < AVB-Threshold and Num-Beta > Num-Beta-Threshold, determines that the user is in an awake and quiet state when AVB > AVB-Threshold and Num-Alpha > Num-Alpha-Threshold, and determines that the user is in an attention-transfer state otherwise.

Different parameters are used as judgment bases on each node of the decision tree, and the selected parameters can effectively distinguish the sleep states of the users, so that the judgment process is effectively simplified, the judgment efficiency is improved, and the classification accuracy is also ensured.

Preferably, at the same node, there is data that can alternatively and independently implement sleep state classification, in particular: setting a Threshold Num-LCZ-Threshold at a first node, when Num-LCZ is more than Num-LCZ-Threshold, judging the single-frame data packet as an artificial artifact, and outputting a sleep state obtained when the previous single-frame data packet is processed to the outside through a processor, otherwise, switching to a second node; or setting a Threshold BVS-Threshold at the second node, and when the BVS is more than the BVS-Threshold, the processor judges that the user is in a sleep latency period or a rapid eye movement period and shifts to the fifth node, otherwise, the processor judges that the user is in a light sleep period, a light sleep period or a deep sleep period and shifts to the third node; or, at the fifth node, setting thresholds TVB-Threshold and Num-EOG-Threshold, when TVB > TVA-Threshold and Num-EOG > Num-EOG-Threshold, the processor judges that the user is in the fast eye movement period, the system stops working, and records the single frame data packet as the data packet of the falling-asleep frame, otherwise, the processor judges that the user is in the falling-asleep latency period, and then the sixth node is switched to. The judgment structure is verified by replacing the judgment condition, so that various judgment modes can be selected, and the judgment accuracy is effectively improved.

Preferably, the acquisition time interval between adjacent single-frame data packets is a, a is greater than or equal to 30s and less than or equal to 60s, the single continuous acquisition time required for forming a single-frame data packet is B, B is greater than or equal to a, and B is greater than or equal to 5s and less than or equal to 60s, the data acquisition frequency of the basic data acquirer is C, C is greater than or equal to 200samples/s and less than or equal to 1000samples/s, the data acquisition times of the basic data acquirer when forming a single-frame data packet are n, and n is C B. The acquisition times of the basic data acquisition unit correspond to the formed data quantity, the data obtained after single continuous acquisition form a single-frame data packet, and then the Alpha wave oscillogram, the Beta wave oscillogram, the Theta wave oscillogram and the SEEG wave oscillogram are respectively obtained through the filter. When the quantity of data in a single-frame data packet is more, the more data points used for drawing the waveform chart are increased, so that the waveform chart is more accurately drawn, the drawn waveform chart is more approximate to the graph of a real electric wave, and more accurate data is provided for a processor. When C is less than 200samples/s, the drawn oscillogram has less data points, so that the drawing precision of the oscillogram is influenced; when C is more than 1000samples/s, the drawing precision is improved to a limited extent, but the requirement on hardware is high, more high-frequency interference is introduced, and the performance is lower.

Preferably, the first-order feature value parameter group is obtained by:

firstly, respectively carrying out absolute value processing on numerical values corresponding to each acquisition point in Alpha wave oscillogram, Beta wave oscillogram, Theta wave oscillogram and SEEG wave oscillogram in sequence, and obtaining a data set X_m；

Thereafter, data set X is processed_mCarrying out data smoothing processing and obtaining a transition data set Y_m，The data length of the transition data group is m, m is more than or equal to 1 and less than or equal to n-0.5C, wherein 0.5C is the length of data smoothing processing;

finally, for the transition data set Y_mAnd performing accumulation calculation and sequentially obtaining the Alpha value, the Beta value, the Theta value and the SEEG value.

When the basic data acquisition unit acquires the basic data, a user can have a jitter value which obviously exceeds the variation range due to various factors, so that the instantaneous change of the oscillogram is large, the sleep state of the user can be judged by the processor under the influence of no influence on the sleep state of the user, the continuous 0.5 × C data is subjected to accumulation average processing, and the influence of a single jitter value on the judgment of the sleep state of the user by the processor is effectively reduced. In addition, data set X_mEach data and 0.5 x C data after the data are accumulated and averaged to form a transition data group Y_mThe number of data m is reduced to n-0.5C.

Preferably, each parameter in the primary characteristic value parameter group is divided by the parameter m to obtain a corresponding Mean-Alpha value, Mean-Beta value, Mean-Theta value and Mean-SEEG value, and then AVB value, AVS value, BVS value, TVA value, TVB value and TVS value are obtained by calculation, wherein AVB is Mean-Alpha/Mean-Beta, AVS is Mean-Alpha/Mean-SEEG, BVS is Mean-Beta/Mean-SEEG, TVA is Mean-Theta/Mean-Alpha, TVB is Mean-Theta/Mean-Beta, and TVS is Mean-Theta/Mean-SEEG. The parameters are obtained through accumulation and averaging, and the influence of the jumping parameters on the accuracy of judging the sleep state of the user is effectively reduced.

Preferably, a corresponding maximum threshold value and a corresponding minimum threshold value are set for each oscillogram, a data fluctuation range is formed according to the maximum threshold value and the minimum threshold value, the number of acquisition points of each oscillogram, which exceed the data fluctuation range, is accumulated and counted, and a Num-ARI value is formed, wherein the Num-ARI value is a parameter for counting the number of jitter values exceeding a preset range.

Preferably, the times of the curve crossing the x axis in each oscillogram are accumulated and counted to form a Num-LCZ value, the times of the curve crossing the x axis in the oscillogram are related to the waveform frequency, and the parameter Num-LCZ value is used for counting the waveform frequency.

Preferably, the duration of the Alpha wave in the Alpha wave waveform diagram is counted to obtain a Num-Alpha value for representing the intensity of the Alpha wave.

Preferably, the time length of the Beta wave appearing in the Beta wave oscillogram is counted, so that a Num-Beta value is obtained, and the Num-Beta value is used for representing the strength of the Beta wave.

Preferably, the occurrence duration of the Theta wave in the Theta wave oscillogram is counted to obtain a Num-Theta value for reflecting the intensity of the Theta wave.

Preferably, the eye movement threshold is set, the eye electrical pulse signal is obtained from the SEEG wave oscillogram, the times that the amplitude of the eye electrical pulse signal exceeds the eye movement threshold is counted, the Num-EOG value is obtained according to the times, and the eye movement pulse signal is obtained through the following steps:

firstly, counting the values of each acquisition point of the SEEG wave oscillogram to form a basic value group with n number;

sequentially extracting X window value groups from the basic value group, wherein the window value groups comprise L basic values sequentially intercepted along the basic value group, X is n-L +1, L is less than n, and L is an odd number;

then, L basic values in each window value group are sequenced from small to large, and the (L +1)/2 th basic value is set as a smooth value of the window value group;

then, forming a smooth value group by the smooth values obtained in each window value group, wherein the smooth value group comprises X smooth values;

and finally, subtracting the numerical values of the corresponding sequence in the smooth numerical value set from the X numerical values in the middle of the basic numerical value set to form X difference values, wherein the difference values form the eye electrical pulse signals.

Preferably, the device for the sleep-onset latency refined classification system comprises a basic data collector and a processor. The basic data collector is used for obtaining basic data through the portable wearable equipment and is provided with an electric signal collecting assembly; and the processor receives the data from the basic data acquisition unit and calculates the sleep state subdivision when the user is in the sleep latency period. The basic data collector and the processor are both arranged on the same device, so that the structure is simplified, the system volume is effectively reduced, and the carrying and the use are convenient.

Preferably, the basic data collector is glasses, the glasses comprise a glasses frame and glasses legs, and the electric signal collecting assembly comprises a nose bridge reference electrode arranged in the middle of the glasses frame and a left front temporal electrode and a right front temporal electrode arranged at two ends of the glasses frame. When the glasses are worn in place, the left front temporal electrode, the right front temporal electrode and the nose bridge reference electrode synchronously collide with a station to be detected respectively, so that the basic data acquisition unit can continuously and accurately acquire data, and a single-frame data packet is periodically sent to the processor.

Preferably, the basic data collector is an eye mask, the eye mask comprises a mask body and a binding band, and the electric signal collecting assembly comprises a frontal reference electrode arranged in the middle of the mask body, and a left frontal electrode and a right frontal electrode arranged at two ends of the mask body. The eye cover can not only play a role in shielding light, but also provide comfortable feeling for users.

The invention has the beneficial effects that: this system has the characteristics that use basic data kind few, calculation and classification process are simple, has both simplified basic data collection station's structure through reducing basic data collection kind, and convenience of customers dresses the use, promotes the data accuracy, still reduces the requirement to the treater through simplifying calculation and categorised process, promotes the arithmetic speed, reduces the hardware cost, ensures hardware operating stability, promotes and uses experience.

Drawings

FIG. 1 is a schematic view of the structure of the glasses;

FIG. 2 is a schematic diagram of the decision tree structure;

FIG. 3 is a schematic view of the eye mask;

in the figure: 1. glasses, 2, a glasses frame, 3, glasses legs, 4, a left front temporal electrode, 5, a right front temporal electrode, 6, a nose bridge reference electrode, 7, an eye patch, 8, a left frontal electrode, 9, a right frontal electrode, 10 and a frontal reference electrode.

Detailed Description

The essential features of the invention will be further explained below with reference to the drawings and the detailed description of the specification.

As shown in fig. 1, in the system for refining and classifying sleep-in latency, a processor acquires basic data of a user through a basic data acquisition unit, obtains a refined sleep state of the user in the sleep-in latency through calculation, and provides a reference basis for implementing the operation of sleep-assisting equipment.

In this embodiment, the device for implementing the method includes a basic data collector and a processor. The basic data collector is used for obtaining basic data through the portable wearable equipment and is provided with an electric signal collecting assembly; and the processor receives the data from the basic data acquisition unit and calculates the sleep state subdivision when the user is in the sleep latency period.

In this embodiment, basic data collector is for taking glasses 1 of data collection subassembly, glasses 1 includes mirror holder 2 and locates the mirror leg 3 at 2 both ends of mirror holder, the data collection subassembly is including putting separately left front temporal electrode 4, right front temporal electrode 5 and locate mirror holder 2 middle part bridge of the nose reference electrode 6 at 2 both ends of mirror holder, glasses 1 is dressed the back in place, left front temporal electrode 4, right front temporal electrode 5 and bridge of the nose reference electrode 6 contradict respectively on waiting to detect the station. The glasses 1 collect relevant basic data through a left front temporal electrode 4, a right front temporal electrode 5 and a nose bridge reference electrode 6 which are attached to the body surface of a user, and transmit the relevant basic data to a processor. Specifically, the left front temporal electrode 4, the right front temporal electrode 5 and the nose bridge reference electrode 6 are respectively attached to the near left front temporal electrode, the near right front temporal electrode and the nose bridge, so that a single lead collector for simply measuring electroencephalogram activity is formed. The basic data acquisition unit acquires Alpha waves, Beta waves, Theta waves and SEEG waves, wherein the Alpha waves, the Beta waves and the Theta waves are brain waves, and the SEEG waves are muscle waves.

In this embodiment, the system is implemented by the following steps when running:

the method comprises the following steps that firstly, basic data are obtained through a basic data collector, in the collection process, the basic data are obtained through an interval collection mode by the basic data collector, and the basic data collected continuously at a time form a single-frame data packet.

Specifically, the acquisition time interval between adjacent single-frame data packets is a, a is 30s, and the single continuous acquisition time required for forming a single-frame data packet is B, B is 10 s. Specifically, the whole sleep process of the user is divided into 30s intervals, and 10s of data acquisition operation is carried out in each interval, so as to form a single-frame data packet. The acquisition times of the basic data acquisition unit correspond to the formed data quantity, the data obtained after single continuous acquisition form a single-frame data packet, and when the data quantity in the single-frame data packet is more, the more data points are used for drawing a waveform diagram, so that the waveform diagram is more accurately drawn, the more the drawn waveform diagram tends to the graph of a real electric wave, and more accurate data is provided for a processor.

And secondly, filtering the single-frame data packet by the processor, and sequentially obtaining an Alpha wave oscillogram, a Beta wave oscillogram, a Theta wave oscillogram and an SEEG wave oscillogram with frequencies of 8-12 Hz, 18-30 Hz, 4-7 Hz and 40-50 Hz respectively.

Specifically, the data collection frequency of the basic data collector is C, C is 250samples/s, preferably, the number of data collection times performed by the basic data collector when a single frame data packet is formed is n, n is C, B is 250samples/s, 10s is 2500, and when the parameter n is a non-integer, the data collection is rounded by rounding. The single-frame data packet acquired by the basic data acquisition unit each time is comprehensive data, and the comprehensive data is subjected to filtering decomposition by using a filter to obtain an Alpha wave oscillogram, a Beta wave oscillogram, a Theta wave oscillogram and a SEEG wave oscillogram with differentiated frequencies. Each waveform has 2500 base points formed in series along a time sequence. The EEG signal in a specific frequency band is extracted by denoising the single-frame data packet preferably by using an 8-order Butterworth filter, a Chebyshev filter or a 4-order Butterworth filter.

And thirdly, counting the numerical values of the acquisition points in each oscillogram, and sequentially carrying out absolute value processing and data smoothing processing, and then obtaining a primary characteristic value parameter group and an auxiliary characteristic value parameter group through accumulation calculation, wherein the primary characteristic value parameter group comprises an Alpha value, a Beta value, a Theta value and a SEEG value, and the auxiliary characteristic value parameter group comprises a Num-ARI value, a Num-LCZ value, a Num-Alpha value, a Num-Beta value, a Num-Theta value and a Num-EOG value.

Specifically, the first-level feature value parameter group is obtained by the following steps:

firstly, respectively carrying out absolute value processing on numerical values corresponding to each acquisition point in Alpha wave oscillogram, Beta wave oscillogram, Theta wave oscillogram and SEEG wave oscillogram in sequence, and obtaining a data set X_m；

Thereafter, data set X is processed_mCarrying out data smoothing processing and obtaining a transition data set Y_m，The data length of the transition data group is m, m is more than or equal to 1 and less than or equal to n-0.5C, wherein 0.5C is the length of data smoothing processing;

finally, for the transition data set Y_mAnd performing accumulation calculation and sequentially obtaining the Alpha value, the Beta value, the Theta value and the SEEG value.

Because Alpha wave oscillogram, Beta wave oscillogram, Theta wave oscillogram and SEEG wave oscillogram are all 25Since 00 base points are connected, the corresponding Alpha value, Beta value, Theta value, and SEEG value are calculated from the 2500 base points corresponding to each waveform diagram. In practical operation, taking an Alpha wave waveform diagram as an example, the Alpha wave waveform diagram includes 2500 base points, and during calculation, first, absolute value processing is performed on the numerical values of the 2500 base points, so as to obtain a data set X including 2500 data_m(ii) a Thereafter, data set X is processed_mCarrying out data smoothing processing and obtaining a transition data set Y_mWhen the treatment is carried out, the treatment,the length of the data smoothing process is 0.5C, the data length of the transition data group is m, m-n-0.5C 2375, and the data length of the data group X is equal to m, n-0.5C 2375_mWhen the 2376 th data is calculated, the number of the data is less than 125, so that the transition data group Y_mThe data length of (2) is 2375; finally, for the transition data set Y_mThe 2375 values are accumulated and the Alpha value is obtained. And so on to obtain Beta, Theta, and SEEG values. When the length of data smoothing is 0.5C is a non-integer, rounding is performed by rounding.

Specifically, a corresponding maximum threshold value and a corresponding minimum threshold value are set for each oscillogram, a data fluctuation range is formed according to the maximum threshold value and the minimum threshold value, the number of acquisition points of each oscillogram, which exceed the data fluctuation range, is accumulated and counted, and a Num-ARI value is formed.

Specifically, the times of the curve in each oscillogram crossing the x axis are accumulated and counted to form a Num-LCZ value.

Specifically, in the whole sleep process of the user, the Alpha wave, the Beta wave, the Theta wave and the EEG wave are individually lost due to different sleep states of the user, and when the basic data acquisition unit cannot acquire a signal, the base point value corresponding to the waveform at the moment is 0, so that a corresponding waveform diagram cannot be formed, and further, the corresponding auxiliary characteristic value parameter is conveniently calculated. And counting the occurrence time of Alpha waves in the Alpha wave oscillogram so as to obtain the Num-Alpha value. And counting the occurrence time of the Beta wave in the Beta wave oscillogram so as to obtain the Num-Beta value. And counting the occurrence time of the Theta wave in the Theta wave oscillogram so as to obtain the Num-Theta value.

Specifically, an eye movement threshold value is set, an eye electrical pulse signal is obtained from an SEEG wave oscillogram, the times that the amplitude of the eye electrical pulse signal exceeds the eye movement threshold value are counted, a Num-EOG value is obtained through the statistics, the eye movement pulse signal is obtained through the following steps, and the example that a single-frame data packet comprises 2500 acquisition points is taken as an example:

firstly, counting the values of each acquisition point of the SEEG wave oscillogram to form n basic value groups, wherein when a single-frame data packet contains 2500 acquisition points, n is 2500, and thus forming the basic value groups containing 2500 values;

then, sequentially extracting X window value groups from the basic value group, wherein the window value groups comprise L basic values which are sequentially intercepted along the basic value group, wherein X is n-L +1, L is less than n, and L is an odd number, sequentially sorting the 2500 values, assuming that the number of values L included in the window value groups is 125, and the number of the obtained window value groups is X2500 + 125+1 is 2376, grouping the basic values numbered from 1 st to 125 th in the basic value group into a 1 st window value group, grouping the basic values numbered from 2 nd to 126 th in the basic value group into a 2 nd window value group, and so on, grouping the basic values numbered from 2376 th to 2500 th in the basic value group into 2376 th window value group;

then, L base values in each window value group are sorted from small to large, and the (L +1)/2 th base value is set as the smoothed value of the window value group, and if L is 125, the 63 rd base value is set as the smoothed value of the window value group. Specifically, 125 basic values in the 1 st window value group are rearranged from small to large, the 63 st basic setting is set as the smooth value of the 1 st window value group, and so on, the smooth values of the 2376 window value groups are respectively obtained;

then, the smooth values obtained in each window value group form a smooth value group, the smooth value group comprises X smooth values, and the statistics of 2376 obtained smooth values are put into the smooth value group;

finally, subtracting the values in the corresponding sequence in the smooth value group from the X values in the middle of the basic value group to form X differences, where the differences constitute the ocular electrical pulse signal, and the X values in the middle of the basic value group are the X values located in the middle of the basic value group, assuming that the basic value group includes 2500 basic values and X equals 2376, the interception is performed in a manner of removing 62 basic values from the front to the back and 62 basic values from the back to the front in the basic value group, thereby obtaining the basic value group including 2376 basic values. And sequentially and respectively subtracting the numerical values of the corresponding numbers in the smooth numerical value group from the 2376 basic numerical values in the basic numerical value group, thereby obtaining the eye electrical pulse signal containing 2376 difference values.

And fourthly, calculating to obtain a secondary characteristic value parameter group through the primary characteristic value parameter group, wherein the secondary characteristic value parameter group comprises an AVB value, an AVS value, a BVS value, a TVA value, a TVB value and a TVS value.

Specifically, each parameter in the primary characteristic value parameter group is divided by the parameter m to obtain a corresponding Mean-Alpha value, Mean-Beta value, Mean-Theta value and Mean-SEEG value, and then an AVB value, an AVS value, a BVS value, a TVA value, a TVB value and a TVS value are obtained through calculation. The Mean-Alpha value, Mean-Beta value, Mean-Theta value, and Mean-SEEG value are averages of base point values of waveforms in a single frame of data packets, for example, the Mean-Alpha value is obtained by dividing the Alpha value by the parameter m 2375, and so on.

Specifically, the AVB is Mean-Alpha/Mean-Beta and is used for representing the strong and weak contrast between an Alpha wave signal and a Beta wave signal in a single frame data packet, when the parameter AVB value is larger, the Alpha wave signal is stronger than the Beta wave signal, and conversely, when the parameter AVB value is smaller, the Alpha wave signal is weaker than the Beta wave signal.

Specifically, the AVS is Mean-Alpha/Mean-seg and is used for representing the amplitude ratio of the Alpha wave signal in the single-frame data packet in the total brain wave signal seg, and when the parameter AVS value is larger, it is indicated that the amplitude ratio of the Alpha wave signal in the total brain wave signal seg is larger, whereas when the parameter AVS value is smaller, it is indicated that the amplitude ratio of the Alpha wave signal in the total brain wave signal seg is smaller.

Specifically, the BVS is Mean-Beta/Mean-seg and is used for indicating the amplitude ratio of the Beta wave signal in the total electroencephalogram signal seg in a single frame data packet, and when the parameter BVS value is larger, it indicates that the amplitude ratio of the Beta wave signal in the total electroencephalogram signal seg is larger, and conversely, when the parameter BVS value is smaller, it indicates that the amplitude ratio of the Beta wave signal in the total electroencephalogram signal EEG is smaller.

Specifically, the TVA is Mean-Mean/Mean-Alpha and is used to represent a strong-weak comparison between a Theta wave signal and an Alpha wave signal in a single frame data packet, and when the parameter TVA value is larger, it is indicated that the Theta wave signal is stronger than the Alpha wave signal, whereas when the parameter TVA value is smaller, it is indicated that the Theta wave signal is weaker than the Alpha wave signal.

Specifically, the TVB is Mean-Mean/Mean-Beta and is used to represent the strong and weak comparison between the Theta wave signal and the Beta wave signal in the single frame data packet, when the parameter TVB value is larger, the Theta wave signal is stronger than the Beta wave signal, and conversely, when the parameter TVB value is smaller, the Theta wave signal is weaker than the Beta wave signal.

Specifically, the TVS is Mean-Mean/Mean-seg and is used to indicate an amplitude ratio of the Theta wave signal in the single frame data packet in the total brain wave signal seg, where the larger the parameter TVS value is, the larger the amplitude ratio of the Theta wave signal in the total brain wave signal seg is, and the smaller the parameter TVS value is, the smaller the amplitude ratio of the Theta wave signal in the total brain wave signal seg is.

And fifthly, finely classifying the sleep state of the user in the sleep-in latent period through the auxiliary characteristic value parameter group and the secondary characteristic value parameter obtained in the third step and the fourth step, so that the sleep state of the user is classified into a waking active state, an attention transfer state or a waking quiet state. The processor classifies the sleep state of the user through a decision tree, wherein at least six nodes are set on the decision tree to classify the sleep state (as shown in fig. 2), specifically:

setting a Threshold Num-ARI-Threshold at a first node, when Num-ARI is larger than Num-ARI-Threshold, judging the single-frame data packet as an artificial artifact, and outputting a sleep state obtained when the previous single-frame data packet is processed to the outside through a processor, otherwise, switching to a second node;

in particular, artificial artifacts refer to interfering signals that affect electrophysiological signals, including physiological artifacts and device artifacts. When the artificial artifact occurs, the basic data can jump beyond a normal range, whether the artificial artifact is generated is judged by comparing the parameter Num-ARI value with the corresponding Threshold Num-ARI-Threshold, specifically, when Num-ARI is larger than Num-ARI-Threshold, the single-frame data packet is judged as the artificial artifact, the single-frame data packet is invalidated, and the sleep state obtained when the previous single-frame data packet is conveyed outwards, otherwise, the manual artifact does not exist at the moment, and further analysis can be carried out.

Setting a Threshold value AVS-Threshold at a second node, when the AVS is larger than the AVS-Threshold, judging that the user is in a sleep latency period or a rapid eye movement period by a processor, and switching to a fifth node, otherwise, judging that the user is in a light sleep period, a light sleep period or a deep sleep period by the processor, and switching to a third node;

specifically, when a user is in a rapid eye movement state or a waking state, the brain is active and mainly emits alpha waves and beta waves, when the user is in a shallow sleep state, the brain mainly emits theta waves, when the user is in a light sleep state, the brain mainly emits K-complex waves and spindle waves with the frequency of 11-16 Hz lasting for several seconds, and when the user is in a deep sleep state, the brain mainly emits delta waves. Therefore, the sleep state of the single-frame data packet can be distinguished as a rapid eye movement period or a waking period, or a light sleep period, a light sleep period or a deep sleep period through the amplitude ratio of the alpha wave and the beta wave in the total brain wave signal. When AVS > AVS-Threshold, it indicates that the amplitude of alpha wave signal or beta wave signal in the single frame data packet is stronger, and it indicates that the brain of the user is active and in REM state or awake state. And conversely, the amplitude ratio of the alpha wave signal and the beta wave signal in the single-frame data packet is weaker, and the brain of the user is inactive and is in a light sleep state, a light sleep state or a deeper sleep state. The K-complex wave includes a sigma wave and a delta wave. spindle waves are sleep spindle waves.

At a third node, setting Threshold values TVS-Threshold and Num-Theta-Threshold, when TVS is larger than TVS-Threshold and Num-Theta is larger than Num-Theta-Threshold, judging that the user is in a light sleep period by a processor, switching to a fourth node, otherwise, judging that the user is in a light sleep period or a deep sleep period by the processor, stopping the system, and recording the single-frame data packet as a data packet of the falling-asleep frame;

specifically, when a user is in a light sleep state, the brain mainly emits theta waves, when the user is in a light sleep state, the brain mainly emits K-complex waves and spindle waves with the frequency of 11-16 Hz and lasting for a plurality of seconds, and when the user is in a deep sleep state, the brain mainly emits delta waves. Therefore, whether the user is in a light sleep state or a deep sleep state can be distinguished through the amplitude ratio condition of the Theta wave signal in the single frame data packet, so that when the TVS is greater than TVS-Threshold and the Num-Theta is greater than the Num-Theta-Threshold, the Theta wave signal is stronger at the moment, which indicates that the brain of the user is inactive and in the light sleep state, and vice versa, which indicates that the brain of the user is active and in the deeper light sleep or deep sleep state. The sleep criteria for MSLT for scientific purposes were: when 3 frames of shallow sleep continuously appear, the sleep starts from the first shallow sleep frame record; any sleep period other than the 1-frame light sleep period, including light sleep period, deep sleep period, and dreaming period, occurs, and the sleep onset starts from the recorded frame. Therefore, when the TVS is more than TVS-Threshold and Num-Theta is more than Num-Theta-Threshold, the T4 node is switched to continue to count the continuous occurrence times of the light sleep period, otherwise, the processor judges that the sleep state of the user is in the light sleep state or the deep sleep state, the sleep detection is finished, and the single-frame data packet is recorded as the data packet of the falling-asleep frame. MSLT is a multiple nap latency check.

At a fourth node, when the processor classifies at least 3 continuous single-frame data packets as a light sleep period, the system stops running, and records the first single-frame data packet classified as the light sleep period as a sleep frame data packet, otherwise, the first node is switched to and processes the next single-frame data packet;

specifically, therefore, when the sleep stage status of the previous 2 consecutive frames is a shallow sleep period, the sleep detection is ended, and the data packet of the falling-asleep frame is determined from the 1 st shallow sleep period in the consecutive 3 frames, otherwise, the detection of the next single-frame data packet is continued.

Setting Threshold values TVA-Threshold and Num-EOG-Threshold at a fifth node, when TVA > TVA-Threshold and Num-EOG > Num-EOG-Threshold, judging that the user is in a fast eye movement period by a processor, stopping the system, recording the single frame data packet as a sleep-in frame data packet, and otherwise, judging that the user is in a sleep-in latent period by the processor and transferring to a sixth node;

specifically, when the user is awake, the brain is active and emits mainly alpha waves and beta waves, the most obvious change at this stage is the appearance of rapid, irregular eye movements when the user is in the rapid eye movement phase, and the brain wave state is similar to that at the time of awake relaxation, mainly embodied as a mixture of alpha, beta and theta waves. Therefore, when TVA > TVA-Threshold and Num-EOG > Num-EOG-Threshold or TVB > TVB-Threshold and Num-EOG > Num-EOG-Threshold, the processor determines that the brain of the user is in a REM state containing theta wave components and irregular eye movement occurs, ends sleep detection, and records the single frame data packet as the data packet of the falling asleep frame. Otherwise, the thinking of the brain of the user is in a waking state, and the T6 node is carried out for further analysis and processing.

And setting thresholds AVB-Threshold, Num-Alpha-Threshold and Num-Beta-Threshold at a sixth node, wherein when AVB is < AVB-Threshold and Num-Beta > Num-Beta-Threshold, the processor determines that the user is in an awake and active state, when AVB is > AVB-Threshold and Num-Alpha > Num-Alpha-Threshold, the processor determines that the user is in an awake and quiet state, otherwise, the processor determines that the user is in an attention transfer state.

Specifically, when a user is awake and thinking is active, the brain emits beta waves with the frequency of 18-30 Hz; when a user is awake and relaxed, the brain emits alpha waves with a frequency between 8 Hz and 12 Hz. Therefore, when AVB < AVB-Threshold, and Num-Beta > Num-Beta-Threshold, it is indicated that the Alpha wave signal is weaker than the Beta wave signal, the user is in a state of waking activity in which the brain is thinking active. When AVB > AVB-Threshold and Num-Alpha > Num-Alpha-Threshold, it is indicated that Alpha wave signal is stronger than Beta wave signal, and the user is in a state of waking silence with quiet brain thinking. Otherwise, it is indicated that the difference between the Alpha wave signal amplitude and the Beta wave signal amplitude is not large, and the user is in a mixed state where the difference between the Alpha wave signal amplitude and the Beta wave signal amplitude is not large, that is, the user is in an attention transfer state.

It is understood that the parameter A can also be 35s, 40s, 50s, 60s, etc., as long as the requirement that A is less than or equal to 30s and less than or equal to 60s is met.

It is understood that the parameter B can also be 5s, 7s, 11s, 20s, 30s, etc., as long as the requirements of B ≦ A and 5s ≦ B ≦ 60s are met.

It is understood that the parameter C can also be 200samples/s, 300samples/s, 500samples/s, 1000samples/s, etc., as long as the requirement of 200samples/s ≦ C ≦ 1000samples/s is met.

It is understood that the parameter m may also be 1, 10, 500, etc., as long as the requirement of 1. ltoreq. m.ltoreq.n-0.5. multidot.C is met.