CN112992379A

CN112992379A - Individual physiological state monitoring and analyzing method and equipment

Info

Publication number: CN112992379A
Application number: CN202010205752.7A
Authority: CN
Inventors: 兰珂; 冯昊翔; 郑捷文; 都昌平
Original assignee: Beijing Haisi Ruige Technology Co ltd
Current assignee: Beijing Haisi Ruige Technology Co ltd
Priority date: 2020-03-23
Filing date: 2020-03-23
Publication date: 2021-06-18

Abstract

The application relates to a personalized physiological state monitoring and analyzing method, which comprises the following steps: finding n index vectors closest to the input vector in an index database G, wherein the n index vectors form a matrix D; by using

Finding an optimized parameter w; obtaining an estimated vector V using an optimization parameter w_est，V_estDw; for each index vector x of matrix D_jFinding the n index vectors closest to the n index vectors and forming a matrix D_jBy using

Finding an optimized parameter w_j(ii) a Using an optimization parameter w_jFinding an estimated vector V_jest，V_jest＝D_jw_j(ii) a Calculating a difference between the input vector and its estimated vector as a first difference V_re(ii) a Calculating each index vector x in the matrix D_jAnd its estimated vector V_jestAs a second difference value; all the second differences form a difference library G_re(ii) a According to the first difference value V_reIn a difference library G_reThe state change index is calculated.

Description

Individual physiological state monitoring and analyzing method and equipment

Technical Field

The invention relates to a technology for monitoring and analyzing human physiological parameters, in particular to a method and equipment for acquiring basic physiological signals of a registration object through wearable equipment, learning continuous physiological signals and further quantitatively analyzing physiological state changes.

Background

The guardian technology has emerged in the last century. From early bedside monitoring, to mobile monitoring, and to current wearable physiological monitoring. In the field of physiological monitoring, there is an important concept of how to characterize changes in physiological state and quantify the changes in physiological state. The traditional method usually relies on human experience to observe the change of absolute value of one or more physiological indexes, and when the observed indexes exceed a certain threshold, an observer can think that the state is abnormal, so as to take the next strain measure. This subjective experience is currently the dominant method in most industries and industries. However, as the system becomes more complex, this method can no longer meet the actual task needs, because the establishment of the threshold value in many problems is accompanied by strong personal subjectivity, and for complex systems, the experience of the individual is often inaccurate; on the other hand, due to the complexity of the system, when the observed index exceeds the actually set threshold value, the system is already in a state of breakdown or a rushing edge, and the subjective experience method cannot play a role of early warning, so that the effect of the method in many practical applications is limited. This is especially true for monitoring the condition of the human body. The human body is a complex system, the organs such as heart, liver, lung and the like are connected with each other in a myriad of ways, the change of the state of the human body is difficult to see if the index of one organ is observed, the individuation difference between people is very large, the fixed threshold value of one person is not suitable for the other person, and the work of a pathological monitoring person is greatly complicated.

Disclosure of Invention

In view of the above problems, the present application aims to provide an individualized physiological status monitoring and analyzing method, which is not based on the physiological parameter monitoring and analyzing technology of the traditional threshold method, but identifies the physiological status change through the physiological time series longitudinal comparison analysis of the monitored object, thereby realizing more sensitive, specific and individualized physiological status monitoring.

The application relates to a personalized physiological state monitoring and analyzing method, which comprises the following steps:

taking a vector formed by the numerical values of k physiological parameters in the same time window as an index vector, wherein k is a natural number greater than 1;

forming an index database G by using a plurality of index vectors of the individual under a normal state;

taking the index vector of the current time window of the individual as an input vector V_in；

In an index database G, n index vectors which are closest to the input vector are found through Euclidean distance or a kernel function, and the n index vectors form a matrix D; by using

Finding an optimized parameter w; where e is a noise vector of n x 1, λ₁W is a vector of n x 1, each element of w is more than or equal to 0, and the sum of each element of w is 1;

obtaining an estimated vector V using an optimization parameter w_est，V_est＝Dw；

For each index vector x of matrix D_jIn the index database, the nearest to the index database is found through Euclidean distance or kernel functionN index vectors and form a matrix D_jBy using

Finding an optimized parameter w_j(ii) a Wherein e_jNoise vector of n x 1, λ_j1As a penalty factor, w_jVector of n x 1, w_jEach element of (1) is not less than 0, and w_jThe sum of the elements of (a) is 1;

using an optimization parameter w_jFinding an estimated vector V_jest，V_jest＝D_jw_j；

Calculating a difference between the input vector and its estimated vector as a first difference V_re；

Calculating each index vector x in the matrix D_jAnd its estimated vector V_jestAs a second difference value; all the second differences form a difference library G_re；

According to the first difference value V_reIn a difference library G_reThe state change index is calculated, and the degree of the current state of the individual deviating from the normal state of the individual is reflected by the state change index.

Preferably, a difference library G is calculated_reMean value of (a)₁Sum variance Σ₁；

Calculating difference library G_reMahalanobis distance squared maximum h:

wherein x is_iIs a difference library G_reThe difference vector of (1);

the lambda value of 0.95 quantile of chi-square distribution with the degree of freedom k is calculated:

wherein f (x) is a chi-squared distribution density function with a degree of freedom k;

and (3) calculating to obtain an approximate real covariance matrix:

∑＝∑₁·h/λ；

calculating the Mahalanobis distance h_re：

h_re＝(V_re-μ₁)^T∑^-1(V_re-μ₁)；

The specific value of SCI is then calculated by the chi-squared distribution cumulative distribution function with degree of freedom k:

preferably, the index database is stored in a computing platform; search of the predetermined number of index vectors, the estimated vector V_estThe first difference value V_reThe calculation of the second difference, the calculation of the first difference in a difference library G_reThe calculation of the distribution in (b) and the calculation of the state change index SCI are both performed on the computing platform.

Preferably, the computing platform is locally located, worn on the individual;

the k physiological parameters are obtained by processing physiological signals sensed by a physiological parameter sensor worn by the individual;

the k physiological parameters are transmitted to the computing platform; the computing platform obtains the indicator vector from the received k physiological parameters.

Preferably, the computing platform is remotely located;

the k physiological parameters are sent to the computing platform through wireless transmission; the computing platform obtains the indicator vector from the received k physiological parameters.

Preferably, the state change index SCI is between 0 and 1, with higher values indicating a greater deviation from the normal state.

Preferably, if the state change index SCI is greater than 0 and less than a predetermined value, the input vector is included as one index vector of the index database.

The application relates to an individualized physiological state monitoring and analysis device, comprising: a computing platform;

the computing platform comprises an index database unit, an estimation unit, a difference unit and a state change index unit;

the computing platform is configured to form a vector by using the values of the k physiological parameters of the individual in the same time window as an index vector;

the index database unit uses a plurality of index vectors of the individual under a normal state to form an index database G;

the estimation unit is used for calculating an estimation vector corresponding to the index vector input to the estimation unit;

the difference unit is used for calculating the difference between the two index vectors;

the state change index unit is used for calculating a state change index SCI reflecting the degree of the current state of the individual deviating from the normal state of the individual;

the computing platform takes the index vector of the current time window of the individual as an input vector V_inBased on the input vector V_inSelecting the input vector V from the index database G of the index database unit_inA predetermined number of the index vectors that are closest; the index vectors with the preset number form a matrix D;

estimation unit utilization

the estimation unit uses the optimization parameter w to obtain an estimation vector V_est，V_est＝Dw；

For each index vector x of matrix D_jIn the index database G of the index database unit, the computing platform finds n index vectors closest to the index vectors and forms a matrix D_j；

Estimation unit utilization

the estimation unit uses the optimized parameter w_jFinding an estimated vector V_jest，V_jest＝D_jw_j；

A difference unit calculates a difference between the input vector and its estimated vector as a first difference V_re；

Each index vector x in the difference unit calculation matrix D_jAnd its estimated vector V_jestAs a second difference value; all the second differences form a difference library G_re；

The state change index unit is based on the first difference V_reIn a difference library G_reThe distribution of (1) calculating the state change index SCI, the passing stateThe change index SCI reflects the extent to which the current state of the individual deviates from its normal state.

Preferably, the state change index unit calculates the difference library G_reMean value of (a)₁Sum variance Σ₁；

State change index unit calculation difference library G_reMahalanobis distance squared maximum h:

wherein x is_iIs a difference library G_reThe difference vector of (1);

the state change index unit calculates the lambda value of 0.95 quantile of chi-square distribution with the degree of freedom k:

the state change index unit calculates to obtain an approximate real covariance matrix:

∑＝∑₁·h/λ；

the state change index unit calculates the Mahalanobis distance h_re：

h_re＝(V_re-μ₁)^T∑^-1(V_re-μ₁)；

Then the state change index unit calculates the specific value of SCI by the chi-square distribution cumulative distribution function with the degree of freedom k:

preferably, the device is worn on the individual; the k physiological parameters are obtained by processing physiological signals sensed by a physiological parameter sensor worn by the individual.

Preferably, the computing platform is remotely located; the k physiological parameters are obtained by processing physiological signals sensed by a physiological parameter sensor worn by the individual; and the k physiological parameters are transmitted to the computing platform in a wireless mode.

Compared with the traditional subjective experience threshold method, the method considers the similarity of the front and back states of a single individual and investigates the cooperative similarity relation among multiple indexes, so that the problem of large difference among individuals is solved. On the other hand, even if some single indexes are still in the normal fluctuation range, the physiological state change condition can be predicted more early.

Drawings

FIG. 1 is a flow chart of wearable system data acquisition and transmission;

FIG. 2 is a wearable system and physiological signals acquired by the wearable system;

FIG. 3 is a graph of heart rate, respiration rate, and three-axis acceleration signals with a 30s sliding time window;

FIG. 4 is a flowchart of the overall physiological state monitoring analysis;

FIG. 5a is a diagram illustrating experimental results of a simulated high altitude hypoxic environment according to the first embodiment;

FIG. 5b is a graph showing the experimental results of the simulated high altitude hypoxic environment in the second embodiment;

fig. 6 is a patient's case for emergency atrial fibrillation rescue in the third embodiment.

Detailed Description

The present application will be described in detail below with reference to the accompanying drawings.

Firstly, data acquisition is carried out by wearing a corset on a monitored object, the worn corset comprises an electrocardio lead interface, a respiration sensor, an acceleration sensor and a blood oxygen saturation acquisition device, electrocardio signals, respiration signals, body position physical movement signals and blood oxygen saturation can be acquired respectively, and the acquired signals are integrated into a signal acquisition box. The data sending module can be used for carrying out central computing on the collected data, and the central computing platform can be local or remote. The signal flow diagram is shown in fig. 1.

The wearable system can be a vest or a chest strap, electrocardio, respiration and three-axis acceleration sensor signals are collected, and the electrocardio signals and the respiration signals are further processed to obtain a heart rate and respiration rate time sequence. The wearable system and its acquired raw physiological signals (electrocardiogram, respiration, triaxial acceleration) and heart rate and respiration rate time series are shown in fig. 2.

And sequentially acting on various original signal indexes of the original heart rate and respiration rate signals through smooth filtering, and extracting a heart rate index, a respiration rate index and a triaxial acceleration index of a wearer from the original heart rate and respiration rate signals. Two-dimensional vectors are formed by heart rate and respiratory rate median in a 30s time window (black rectangular frame in fig. 3), current vital signs of a tester are represented, and vectors of a monitored person in a healthy and stable state are usually selected to construct a physiological index state vector library.

After the data is transmitted to the central computing platform, the central computing platform provides independent computing resources for each monitored person, and an independent physiological index state base is established by monitoring physiological parameters for a period of time and is represented by G. After the establishment of G is completed, individual physiological state monitoring analysis can be carried out, and the input index vector is set as V_inGiven a V_inThen, the system will find V from G_inThe most similar 15 vectors form a D matrix, and the degree of similarity can be characterized by euclidean distance, kernel function, etc., where ≦ indicates kernel function operation,

two vectors x are described_iAnd x_jThe degree of closeness is given by the following formula:

wherein the kernel function

Legal kernel functions such as gaussian kernel and trigonometric function kernel can be used.

In finding the optimal estimation vector V_estIn time, the present application employs three methods.

Method one, after obtaining matrix D, applying optimization theory to search V_inOptimal estimation vector V_estThe problem of (2) is converted into an optimization problem of the pre-solved parameter w:

w≥0

wherein, the matrix E uses frobenius norm, D is AND V_inThe matrix of the closest 15 vectors, E being a noise matrix in the form of 15 x 15, E being a noise vector in the form of 15 x 1, w being a coefficient vector of 15 x 1.

In the second method, further regularization is performed on the parameter w, for example, for sparsification of the coefficient w, we will convert the problem of finding the most similar vector into the following optimization problem:

w≥0

wherein, the matrix E uses frobenius norm, the coefficient w vector uses L1 norm, lambda₁Is a penalty term coefficient.

Method three, the noise part can also be simplified, only considering the noise vector e:

w≥0

where e is a noise vector in the form of 15 x 1 and the norm of the coefficient w uses the L2 norm.

After obtaining the coefficient w by the above method, the sum V can be obtained_inIs estimated to be the optimal estimation vector V_est：

v_est＝Dw

Then consider V_estAnd V_inDifference value V of_re：

v_re＝v_in-c_est

Through V_reCan reflect the similarity between the two vectors to describe whether the state is changed or not, when the input vector V is_inVector V selected from computing system_estSimilarly, the current physiological state is considered not to change much, and if the difference between the input vector and the vector selected by the computing system is large, the current physiological state can be considered to change.

Except for the observation of V_reThe value of the value reflects the change of the physiological state, and the more stable method is to observe V_reDistribution of (2).

Further, based on changes in a plurality of physiological parameters, an estimation can be made of a State Change Index (SCI).

When processing a template library G sample, in order to eliminate the influence of abnormal vectors on a result, only sample points in an equal probability range corresponding to 95% quantile points are reserved, so that a calculated residual is no longer a random sample, and an element absolute value of an estimated covariance matrix is small, so that an approximate real covariance matrix needs to be deduced.

Residual G by sample G_reCalculating the mean and variance μ₁，∑₁。

Calculating the maximum squared mahalanobis distance:

wherein x is_iIs G_reThe sample residual vector of (1). And (3) calculating the lambda value of a chi-square distribution 0.95 quantile with the degree of freedom of the characteristic dimension of the sample:

where f (x) is the chi-squared distribution density function with degrees of freedom as the characteristic dimension of the sample.

And (3) calculating to obtain an approximate real covariance matrix:

∑＝∑₁·h/λ

according to the currently calculated residual V_reIn the sample G_reResidual distribution (μ)₁Σ) determines the magnitude of the state change index, the closer to 1 the farther from the state index library G.

First, the mahalanobis distance is calculated:

h_re＝(V_re-μ₁)^T∑^-1(V_re-μ₁)

then, calculating a specific numerical value of SCI by a chi-square distribution cumulative distribution function with the degree of freedom as a sample characteristic dimension k:

at this time, it can be based on V_reIn case of deviation, further selecting V_inAnd adding the state vector into a physiological index state vector library to serve as a new state of the system to enrich a state index library G.

One advantage of the computing platform is that when new state index data V is obtained_inWhen the system comes, the system can be updated regularly, so that the system can automatically monitor for a long time and give early warning in time.

Firstly, in order to verify the accuracy and feasibility of the algorithm, a simulated plateau experiment is carried out on healthy people, physiological parameters of a tester in one day of daily life are used as a state vector library G in the experiment, in a simulated plateau hypoxia environment, the tester can respectively conduct three actions of static reading, fast walking and leg lifting under a simulated environment with the altitude of four kilometers, the activity intensity close to the previous day is kept as much as possible, and whether the current physiological signs of the tester deviate from the state vector library G or not can be verified accurately.

Example one

FIG. 5a is a graph of the results of a simulated high altitude hypoxia experiment performed by a subject in order to better observe the hypoxic environment of the subject in which the subject is located, according to a first method.

At the very beginning of the experiment, the tester was still at normal altitude with a current state vector of V_in＝[73.2602，25.0000]And through similarity comparison, taking out 15 vectors from G to form a matrix D:

heart rate	Respiration rate
		73.2602	25
73.7101	25
		72.3765	25
74.4417	25
		74.4417	25
72.0289	25
		71.8572	25
71.8564	25
		71.6847	25
71.3445	25
		75.2824	25
73.3496	26
		709220	25
72.9931	24
		73.6206	26

The final calculated optimal estimated value V_est＝[73.9942，25.0637]，

Difference V_re＝[0.3160，0.0637]Calculating V_reAt G_reDistribution of

The SCI index in (1) is 0.1584, and the deviation from the state vector library G is small. Instead, we take the vector after the plateau experiment starts: v_in＝[92.7359，22.0000]The corresponding D matrix is:

heart rate	Respiration rate
		76.9231	23
79.1557	24
		77.7202	23.5
78.7402	24
		76.4332	23
80.0000	25
		77.6208	24
74.0741	22.5
		72.8160	22
72.5523	22
		81.3032	26
76.6295	24
		71.6855	22
78.3290	25
		73.7101	21

The final calculated optimal estimated value V_est＝[77.7888，24.0335]，

Difference V_re＝[14.9471，2.0335]The calculated SCI index reached 0.98, deviating significantly from the state vector library G. SCI index is between 0 and 1, with higher values indicating a greater deviation from daily status. As can be seen from FIG. 5a, the physiological status of the subject under the hypoxic environment changes (blood oxygen saturation is decreased), the SCI index is also increased significantly, and the change of the physiological status of the subject can be effectively represented.

Example two

FIG. 5b is a graph illustrating the results of a simulated high altitude hypoxia experiment performed by a subject, wherein FIG. 5b is performed according to a second method in which a blood oxygenation signal is applied to better observe the hypoxic environment of the subject.

At the very beginning of the experiment, the tester was still at normal altitude with a current state vector of V_in＝[68.1818，22.0000]And through similarity comparison, taking out 15 vectors from G to form a matrix D:

heart rate	Respiration rate
		68.3766	22
67.7966	22
		68.9655	22
67.2316	22
		67.2274	22
69.1648	22
		67.0391	22
67.0391	22
		68.1818	23
68.1818	21
		68.3766	23
65.9341	22
		67.7966	21
67.7966	21
		65.9341	22

The final calculated optimal estimated value V_est＝[67.8312，21.9314]，

Difference V_re＝[0.3505，0.0685]Calculating V_reAt G_reDistribution of

In (1)The SCI index is 0.0397, the deviation from the state vector library G is small. Instead, we take the vector after the plateau experiment starts: v_in＝[96.6186，16]The corresponding D matrix is:

heart rate	Respiration rate
		69.9714	19
75.4717	22
		68.9635	19
72.2892	21
		76.9231	23.5
73.1707	22
		67.0391	19
67.0391	19
		71.0059	21
71.0059	21
		68.5714	20
72.5082	22
		74.3342	23
70.3818	21
		72.2918	22

The final calculated optimal estimated value V_est＝[75.1937，22.7126]，

Difference V_re＝[21.4248，6.7126]The calculated SCI index reached 0.99, deviating significantly from the state vector library G. SCI index is between 0 and 1, with higher values indicating a greater deviation from daily status. As can be seen from FIG. 5b, the physiological status of the subject under the hypoxic environment changes (blood oxygen saturation is decreased), the SCI index is also significantly increased, and the change of the physiological status of the subject can be effectively represented.

EXAMPLE III

Fig. 6 is a patient case for rescue of a sudden severe atrial fibrillation. Method three is implemented in fig. 6. Selecting the physiological data of a patient in a relatively steady state for one day to construct a physiological state vector library G, where (a) in fig. 6 is data of a certain morning where vital signs are steady after rescue treatment, for example, about 52 minutes at 08 am, and the current state vector is: v_in＝[51.2820，22]And through similarity comparison, taking out 15 vectors from G to form a matrix D:

heart rate	Respiration rate
		51.3921	22
51.3921	22
		51.3921	22
51.0638	22
		51.0638	22
50.9556	22
		50.8475	22
51.7241	22
		50.6329	22
51.2821	21
		52.0610	22
50.4202	22
		50.4202	22
51.1729	23
		52.1739	22

The final calculated optimized estimated value V_est＝[51.2051，21.9986]The difference vector is: v_re＝[0.0769，0.0014]Calculate it at G_reDistribution:

the SCI value in (1) is 0.0853, and it can be seen from the figure that most of the SCI values are at a normal level. In FIG. 6 (b), the data collected on the day of atrial fibrillation of the patient is shown, and the current input vector is V at about 19 PM and 40 PM_in＝[72.5，12.5]The corresponding matrix D is:

heart rate	Appealing rate
		51.8361	14
50.6329	13
		51.0638	14
52.4017	16
		50.0000	13
52.2878	16
		50.6329	14
50.6329	14
		49.7923	13
53.5725	18
		49.3868	13
51.8361	16
		50.2092	14
50.2092	14
		50.2092	14

The calculated optimal estimate vector: v_est＝[53.5725，18]Difference value V_re＝[18.9275，5.5]The calculated SCI was as high as 0.99, and it was shown from medical records that the patient did feel physically uncomfortable from afternoon and atrial fibrillation occurred around 21 o 'clock and 50 o' clock, but SCI was consistently at a higher level from 19 o 'clock and 40 o' clock than the normal level.

In the application, the physiological state change is identified through the longitudinal comparison and analysis of the physiological time sequence of the monitored object, so that more sensitive, specific and individualized physiological state monitoring is realized.

The computing platform can be realized by a single chip microcomputer, a DSP, a computer and the like, and the index database unit, the estimation unit, the difference unit and the state change index unit can be functional modules realized by programs on the computing platform.

Unless defined otherwise, all technical and/or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The materials, methods, and examples set forth in this application are illustrative only and not intended to be limiting.

Although the present invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the teachings of this application and yet remain within the scope of this application.

Claims

1. A personalized physiological state monitoring analysis method, comprising:

For each index vector x of matrix D_jIn the index database G, n index vectors closest to the index database G are found through Euclidean distances or kernel functions to form a matrix D_jBy using

Calculating a difference between the input vector and its estimated vector as a secondA difference value V_re；

2. The individualized physiological state monitoring analysis method of claim 1, wherein:

calculating difference library G_reMean value of (a)₁Sum variance Σ₁；

Calculating difference library G_reMahalanobis distance squared maximum h:

wherein x is_iIs a difference library G_reThe difference vector of (1);

and (3) calculating to obtain an approximate real covariance matrix:

∑＝∑₁·h/λ；

calculating the Mahalanobis distance h_re：

h_re＝(V_re-μ₁)^T∑^-1(V_re-μ₁)；

3. the individualized physiological state monitoring analysis method of claim 1, wherein:

the index database is stored in a computing platform; search of the predetermined number of index vectors, the estimated vector V_estThe first difference value V_reThe calculation of the second difference, the calculation of the first difference in a difference library G_reThe calculation of the distribution in (b) and the calculation of the state change index SCI are both performed on the computing platform.

4. The individualized physiological state monitoring analysis method of claim 1, wherein:

the computing platform is located locally and worn on the individual;

5. The individualized physiological state monitoring analysis method of claim 4, wherein:

the computing platform is remotely located;

6. The individualized physiological state monitoring analysis method of claim 1, wherein:

the state change index SCI is between 0 and 1, with higher values indicating a greater deviation from the normal state.

7. The individualized physiological state monitoring analysis method of claim 1, wherein:

if the state change index SCI is greater than 0 and less than a predetermined value, the input vector is included as an index vector of the index database.

8. An individualized physiological state monitoring and analysis device, comprising: a computing platform;

estimation unit utilization

Estimation unit utilization

State change index cell rootAccording to the first difference value V_reIn a difference library G_reThe state change index SCI is calculated, and the degree of the current state of the individual deviating from the normal state is reflected by the state change index SCI.

9. The individualized physiological state monitoring and analysis device of claim 8, wherein:

state change index unit calculation difference library G_reMean value of (a)₁Sum variance Σ₁；

wherein x is_iIs a difference library G_reThe difference vector of (1);

∑＝∑₁·h/λ；

the state change index unit calculates the Mahalanobis distance h_re：

h_re＝(V_re-μ₁)^T∑^-1(V_re-μ₁)；

10. the individualized physiological state monitoring and analysis device of claim 8, wherein:

the device is worn on the individual; the k physiological parameters are obtained by processing physiological signals sensed by a physiological parameter sensor worn by the individual.

11. The individualized physiological state monitoring and analysis device of claim 9, wherein:

the computing platform is remotely located; the k physiological parameters are obtained by processing physiological signals sensed by a physiological parameter sensor worn by the individual; and the k physiological parameters are transmitted to the computing platform in a wireless mode.