阅读说明：本技术 一种慢性病患者健康趋势监测与预警系统 (Health trend monitoring and early warning system for chronic disease patients ) 是由 卢国梁 叶新来 毕新杰 欧德怀 闫鹏 于 2021-08-10 设计创作，主要内容包括：本发明公开了一种慢性病患者健康趋势监测与预警系统,其技术方案为：包括压力采集模块、数据处理模块和预警与报警模块,压力采集模块由若干压力传感器呈矩阵式分布,用于获取表征生命体征参数的多维压力信号；数据处理模块用于将多维压力信号数据融合得到一维动态指标；以多个连续的指标点作为图模型的节点,建立动态图模型；通过动态图模型计算优化指标；预警与报警模块用于比较优化指标与阈值以进行趋势预警；基于假设检验方法进行异常报警。本发明通过模块化设计,能够灵活搭配监测垫,实现生命体征状态监测；基于动态图模型构建模块,能够对监测数据进行实时处理得到优化指标,根据不同类型的数据超限或异常状态给出相应预测与报警。(The invention discloses a system for monitoring and early warning the health trend of a chronic patient, which adopts the technical scheme that: the device comprises a pressure acquisition module, a data processing module and an early warning and alarming module, wherein the pressure acquisition module is distributed by a plurality of pressure sensors in a matrix form and is used for acquiring a multidimensional pressure signal representing vital sign parameters; the data processing module is used for fusing the multidimensional pressure signal data to obtain a one-dimensional dynamic index; establishing a dynamic graph model by taking a plurality of continuous index points as nodes of the graph model; calculating an optimization index through a dynamic graph model; the early warning and alarming module is used for comparing the optimized index with a threshold value to perform trend early warning; and carrying out abnormity alarm based on a hypothesis testing method. According to the invention, through the modular design, the monitoring pad can be flexibly matched to realize the monitoring of the vital sign state; based on the dynamic graph model building module, the monitoring data can be processed in real time to obtain an optimized index, and corresponding prediction and alarm are given according to different types of data overrun or abnormal states.)

1. A chronic patient health trend monitoring and early warning system, comprising:

the pressure acquisition module is distributed by a plurality of pressure sensors in a matrix form and is used for acquiring multidimensional pressure signals representing vital sign parameters;

the data processing module is used for fusing the multidimensional pressure signal data to obtain a one-dimensional dynamic index; establishing a dynamic graph model by taking a plurality of continuous index points as nodes of the graph model; converting the one-dimensional dynamic index into a graph model through a dynamic graph model, and calculating an optimization index based on the graph model;

the early warning and alarming module is used for comparing the optimization index with a threshold value to perform trend early warning; and carrying out abnormity alarm based on a hypothesis testing method.

2. The chronic patient health trend monitoring and pre-warning system of claim 1, wherein the data processing module comprises a dynamic graph model building module;

the dynamic graph model building module is used for taking a plurality of continuous index points as nodes of the graph model, and the index points are connected pairwise to form edges; and calculating the Euclidean distance of each side, and expressing each Euclidean distance as a matrix to obtain a corresponding graph model.

3. A chronic patient health trend monitoring and pre-warning system according to claim 1 or 2, wherein the data processing module comprises an optimization index calculation module;

and the optimization index calculation module is used for obtaining an optimization index according to the deviation degree between the current model and the normal model.

4. A chronic patient health trend monitoring and pre-warning system as claimed in claim 3 wherein the normal model is derived from a median map absorbing normal fluctuations.

5. The system of claim 1, wherein the pre-warning and warning module is configured to provide a health warning when the optimization index value exceeds a threshold value.

6. The system of claim 1 or 5, wherein the early warning and alarming module identifies the abnormal point based on a hypothesis testing method by using an optimization index exceeding a confidence interval as a determination condition.

7. The chronic patient health trend monitoring and early warning system of claim 6, wherein the early warning and alarm module comprises an anomaly cause analysis module;

and the abnormal reason analysis module is used for sending the graph model corresponding to the abnormal point into the K neighbor classifier so as to determine the abnormal reason.

8. The system of claim 7, wherein the category of the sample to be tested is determined by voting according to the category of the K-nearest neighbor sample to determine the cause of the abnormality and to record the occurrence frequency of each condition; and carrying out statistical analysis according to the frequency of the abnormal reasons to obtain the change trend of the chronic disease condition.

9. The system for monitoring and pre-warning the health tendency of the chronic patient according to claim 1, wherein the pressure acquisition module is connected to a pre-processing module, and the pre-processing module is used for converting the analog signal of the pressure acquisition module into a digital signal; the preprocessing module is connected with the data processing module through the data transmission module.

10. The system of claim 1, wherein the pressure sensors are arranged in a matrix on the surface of the base.

Technical Field

The invention relates to the field of health monitoring systems, in particular to a system for monitoring and early warning health trends of chronic disease patients.

Background

At present, for partial chronic patients, such as epilepsy and the like, the onset time is random and unpredictable, and long-time real-time monitoring is needed. The inventor finds that common household monitoring equipment comprises a monitoring bracelet, wearable electrode monitoring equipment, video monitoring equipment and the like, and household monitoring usually needs to wear or bind the equipment, so that the equipment has certain influence on normal life and is relatively poor in comfort; is not suitable for night sleep monitoring. The privacy protection of the monitoring equipment is poor, and a great deal of energy is consumed. Meanwhile, the real-time performance of the monitoring devices is low, and early warning cannot be given to the health trend.

In addition, the existing pressure pad products have few measuring points and poor real-time performance, mainly perform static pressure distribution analysis, rely on visual observation and experience subjective decision, consume a large amount of time and energy, and cannot realize real-time monitoring and early warning.

Disclosure of Invention

Aiming at the defects in the prior art, the invention aims to provide a health trend monitoring and early warning system for chronic patients, the monitoring pad can be flexibly matched through modular design, and the monitored personnel can monitor the vital sign state only by lying or sitting on the monitoring pad; and based on the dynamic graph model building module, the monitoring data can be processed in real time to obtain an optimized index, and corresponding prediction and alarm are given according to different types of data overrun or abnormal states.

In order to achieve the purpose, the invention is realized by the following technical scheme:

the embodiment of the invention provides a system for monitoring and early warning the health trend of a chronic patient, which comprises:

the pressure acquisition module is distributed by a plurality of pressure sensors in a matrix form and is used for acquiring multidimensional pressure signals representing vital sign parameters;

the data processing module is used for fusing the multidimensional pressure signal data to obtain a one-dimensional dynamic index; establishing a dynamic graph model by taking a plurality of continuous index points as nodes of the graph model; converting the one-dimensional dynamic index into a graph model through a dynamic graph model, and calculating an optimization index based on the graph model;

the early warning and alarming module is used for comparing the optimization index with a threshold value to perform trend early warning; and carrying out abnormity alarm based on a hypothesis testing method.

As a further implementation, the data processing module includes a dynamic graph model building module;

the dynamic graph model building module is used for taking a plurality of continuous index points as nodes of the graph model, and the index points are connected pairwise to form edges; and calculating the Euclidean distance of each side, and expressing each Euclidean distance as a matrix to obtain a corresponding graph model.

As a further implementation, the data processing module includes an optimization index calculation module;

and the optimization index calculation module is used for obtaining an optimization index according to the deviation degree between the current model and the normal model.

As a further implementation, the normal model is obtained by absorbing normal fluctuations from a median map.

As a further implementation manner, in the early warning and alarming module, when the optimization index value exceeds a threshold value, a health early warning is given.

As a further implementation manner, the early warning and alarm module confirms the abnormal point by taking the optimization index exceeding the confidence interval as a judgment condition based on a hypothesis testing method.

As a further implementation manner, the early warning and alarming module comprises an abnormal reason analyzing module;

and the abnormal reason analysis module is used for sending the graph model corresponding to the abnormal point into the K neighbor classifier so as to determine the abnormal reason.

As a further implementation mode, determining the category of the detected sample by using a voting method according to the category of the K neighbor sample so as to determine the abnormal reason and record the occurrence frequency of various conditions; and carrying out statistical analysis according to the frequency of the abnormal reasons to obtain the change trend of the chronic disease condition.

As a further implementation manner, the pressure acquisition module is connected with a preprocessing module, and the preprocessing module is used for converting an analog signal of the pressure acquisition module into a digital signal; the preprocessing module is connected with the data processing module through the data transmission module.

As a further implementation, the pressure sensors are arranged on the surface of the base in a matrix form.

The invention has the following beneficial effects:

(1) the pressure acquisition module can be freely and flexibly matched with the monitoring pad through modular design, and monitored personnel can acquire pressure change signals in real time only by lying or sitting on the monitoring pad, so that vital sign states such as body movement information and the like can be monitored, and the real-time monitoring can be ensured.

(2) The upper computer comprises a data processing module and an early warning and alarming module, wherein the data processing module comprises a dynamic graph model building module and an optimization index calculating module, the dynamic graph model building module can obtain a graph model, and the optimization index calculating module is used for obtaining an optimization index according to the deviation degree between a current model and a normal model; the optimization indexes based on the dynamic graph model well solve the problems of non-stationarity and interference of the collected data and ensure the accuracy of detection and identification.

(3) According to the method, health early warning is given out by comparing the optimization index with the threshold value; the early warning and alarming module determines abnormal points based on hypothesis testing so as to give an alarm for the attack of the chronic disease; the early warning and alarming module comprises an abnormal reason analyzing module which determines the abnormal reason based on the K neighbor classifier; the change trend of the chronic disease condition is obtained by recording the occurrence frequency of various conditions and carrying out statistical analysis according to the occurrence frequency of abnormal reasons.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.

FIG. 1 is a block diagram of a system in accordance with one or more embodiments of the invention;

FIG. 2 is a schematic diagram of dynamic graph modeling in accordance with one or more embodiments of the present invention.

Detailed Description

The first embodiment is as follows:

the embodiment provides a system for monitoring and early warning of health tendency of chronic patients, especially suitable for night monitoring, as shown in fig. 1, comprising:

the pressure acquisition module is distributed by a plurality of pressure sensors in a matrix form and is used for acquiring multidimensional pressure signals representing vital sign parameters;

the data processing module is used for fusing the multidimensional pressure signal data to obtain a one-dimensional dynamic index; establishing a dynamic graph model by taking a plurality of continuous index points as nodes of the graph model; converting the one-dimensional dynamic index into a graph model through a dynamic graph model, and calculating an optimization index based on the graph model;

the early warning and alarming module is used for comparing the optimization index with a threshold value to perform trend early warning; and carrying out abnormity alarm based on a hypothesis testing method.

Further, pressure sensor is arranged in the base surface and constitutes pressure acquisition module in the matrix. In the present embodiment, each 64 pressure sensors are uniform on the square base in the form of 8 × 8; of course, in other embodiments, the number of pressure sensors may be adjusted to other values. The pressure acquisition module formed by the pressure sensor supports 384 measurement points at most, and can meet the monitoring of vital signs of human heartbeat, respiration and the like.

Set up a plurality of pressure acquisition modules in monitoring pad, monitored personnel only need lie or sit on monitoring pad, can gather pressure variation signal in real time to acquire vital sign state information.

In this embodiment, the pressure sensor is a piezoelectric film sensor, has the characteristics of high sensitivity, high resolution and quick response, and can be used for monitoring the body motion information in real time.

Furthermore, the pressure acquisition module is connected with the preprocessing module, and the preprocessing module is used for converting the analog signal of the pressure acquisition module into a digital signal. In this embodiment, the preprocessing module adopts an a/D conversion module with an STM32F103 single chip microcomputer as a core; it can be understood that in other embodiments, the model of the single chip of the preprocessing module can be selected.

The preprocessing module is connected with the data processing module through the data transmission module, and an IP address is distributed to each preprocessing module through TCP-IP transmission; the upper computer can freely select and collect data of one or more modules, and the data are transmitted in parallel, so that flexible modular configuration is realized.

The upper computer comprises a data processing module, a display module, a storage module and an early warning and alarming module, wherein the display module is used for displaying the stress distribution state in real time through a stress distribution diagram and inquiring monitoring statistical data including sleep conditions, continuous body movement time, alarming information and the like through the display module; the storage module is used for storing data.

Furthermore, the data processing module comprises a dynamic graph model building module and an optimization index calculation module, wherein the dynamic graph model building module is used for taking a plurality of continuous index points as nodes of the graph model, and the index points are connected pairwise to form edges; and calculating the Euclidean distance of each side, and expressing each Euclidean distance as a matrix to obtain a corresponding graph model.

Further, in the dynamic graph model building module, the modeling process executed by the dynamic graph model building module is shown in fig. 2, and includes:

step 1: and (3) multi-dimensional data fusion, wherein the number of the sensors is assumed to be m, the acquired data is extracted by filtering and denoising through wavelet transform to obtain multi-dimensional original data:

and main data fluctuation information is reserved by adopting methods such as principal component analysis/self-adaptive weighted fusion and the like, and m-dimensional data is fused into a one-dimensional dynamic index: x ═ X₁，x₂，…x_n]。

Step 2: the dynamic graph model is optimized, factors such as non-stationarity and other interferences exist in a one-dimensional index obtained by fusion, false alarm or detection failure is easily caused, the problem of non-linear non-stationarity can be well solved by the dynamic graph model, and meanwhile, the dynamic graph model has a good inhibiting effect on interference noise:

1) regarding the continuous j index points as nodes { v ] of the graph model₁，v₂，…v_jConnected two by two, each group of nodes forms an edge l_{a，b}A, b is belonged to {1, 2.. j }, and the Euclidean distance d of the Euclidean distance is calculated_{a，b}To obtain the segmented graph model G_iAnd expressed as a matrix:

2) the modeled raw metrics are represented as a series of graphical models: x: Γ ═ G₁，G₂，...，G_nAnd using a median mapAbsorbing normal fluctuation:

m (-) is a distance metric, here chosen for the DEWV distance metric, namely:

wherein Δ_{a，b}The calculation is as follows:

further, the optimization index calculation module is used for obtaining an optimization index according to the deviation degree between the current model and the normal model, namely calculating G_n+1Andthe deviation degree between the current model and the normal model is obtained:

further, in the early warning and alarming module, a threshold value y is set and compared with the optimization index, and when the optimization index value gradually approaches to the threshold value y and exceeds the threshold value, health early warning is given out, which indicates that a chronic health problem may occur.

When the chronic diseases such as epilepsy and the like suddenly happen, an alarm needs to be given in time; in the early warning and alarming module, based on a hypothesis testing method, the abnormal point is confirmed by taking the optimization index exceeding the confidence interval as a judgment condition.

In the present embodiment, based on the 3 σ criterion of hypothesis testing, upper and lower control limits are determined for a confidence interval beyond which the optimization indicator will be considered as an abnormal condition:

H₀: and (3) normal: s_n+1∈A

H_A: exception:

wherein A ═ mu_n-3σ_n，μ_n+3σ_n]To a confidence interval, μ_nIs a mean value, σ_nStandard deviation, as follows:

the alarm range can be set individually according to the actual application scene, and can be set and managed in different time intervals, and the alarm mode comprises alarm sound and color reminding (through displaying interface thermodynamic diagrams).

Further, the early warning and alarming module comprises an abnormal reason analyzing module; and the abnormal reason analysis module is used for sending the graph model corresponding to the abnormal point into the K neighbor classifier so as to determine the abnormal reason.

For a series of graph models Γ ═ G obtained in the dynamic graph model building block₁，G₂，...，G_nSending the graph structure of the current abnormal moment into a K neighbor classifier after the abnormality occurs: and performing distance measurement on the tested sample and all the training samples so as to determine K adjacent samples of the tested sample.

The graph distance metric uses a weighted edge distance, M, for graph G and G_wedThe calculation formula of (a) is as follows:

wherein Δ_i，jCalculated from the following formula:

finally, determining the category of the detected sample by using a voting method according to the category of the K neighbor sample, determining abnormal reasons (convulsion, snoring and the like), recording the times of occurrence of various conditions, and finally performing statistical analysis according to the times of occurrence of the abnormal reasons to obtain the change trend (no change/good/bad) of the chronic disease condition.

The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.