Real-time monitoring method of vital signs in the elderly based on wearable devices

Abstract

This application relates to the field of intelligent sensor technology, specifically to a method for real-time monitoring of vital signs in the elderly based on wearable devices. The method includes: collecting vital sign data from elderly users; determining data points for each time point using vital sign data from adjacent time points; determining the probability of anomalies in the vital sign data based on fitted curves; determining the likelihood of anomalies in data points based on the probability of anomalies and the differences between trend terms of different fitted curves; clustering the selected feature data points; classifying abnormal clusters and normal clusters based on the clustering characteristics and intra-cluster distribution of feature data points within different clusters; determining the correction result of the vital sign data based on the values ​​of feature data points within different clusters; and obtaining the real-time monitoring result of the vital sign data. This application improves the accuracy of vital sign monitoring data by analyzing the differences in vital sign data caused by abnormal states of elderly users and movement of wearable devices to correct the data.

Description

Technical Field

[0001] This application relates to the field of intelligent sensor technology, specifically to a method for real-time monitoring of vital signs in the elderly based on wearable devices. Background Technology

[0002] As people's living standards improve, they are paying more and more attention to their health, making it particularly important to obtain health status parameters through various devices.

[0003] With the widespread adoption of wearable devices, such as smartwatches, wristbands, and vests, which utilize built-in biosensors to collect basic physiological indicators like heart rate, blood pressure, and body temperature in real time, wearable devices offer excellent portability and ease of use, making them popular in the elderly care and health sector. Wearable devices combine data collection, transmission, and analysis through intelligent sensors to provide 24 / 7 health protection for the elderly.

[0004] When using wearable devices to monitor the health of the elderly in real time, the vital signs data collected by the built-in sensors, after analysis and visualization, helps the elderly to have a preliminary understanding of their own condition and also assists elderly care medical staff in conducting more professional monitoring and analysis. Vital signs monitoring data from wearable devices, such as heart rate, are generally obtained by collecting and calculating heart rate data from blood flow within the skin at the wearing site. However, due to factors such as potential displacement of the wearable device during the elderly's movement or poor contact between the sensor and the skin, the collected data may be inaccurate, leading to certain deviations in the vital signs monitoring results. Summary of the Invention

[0005] To address the issue of inaccuracies in existing methods for calculating human heart rate, this application aims to provide a real-time monitoring method for vital signs in the elderly based on wearable devices. The specific technical solution adopted is as follows:

[0006] This application provides a method for real-time monitoring of vital signs in the elderly based on wearable devices, the method comprising the following steps:

[0007] Collect vital sign data of elderly users using wearable devices;

[0008] The data points corresponding to each time step in the multidimensional data space are determined by using vital sign data collected at adjacent time steps;

[0009] The abnormal probability of each type of vital sign data at each time point is determined based on the fitting curve of the vital sign data collected at each time point and at historical time points. The abnormal probability of the corresponding data point at each time point is determined based on the difference between the trend terms of the fitting curves of different types of vital sign data at each time point and the abnormal probability of the data point at each time point.

[0010] Cluster the feature data points for the anomaly probability screening, and divide the abnormal clusters and normal clusters based on the clustering characteristics and intra-cluster distribution of the feature data points within different clusters in the clustering results.

[0011] Based on the values ​​of feature data points within abnormal and normal clusters, the correction results of vital sign data at the corresponding time for each feature data point within the abnormal cluster are determined, and the real-time monitoring results of vital sign data for elderly users are obtained based on the correction results.

[0012] Preferably, determining the data point corresponding to each time moment in the multidimensional data space includes:

[0013] The ratio of the absolute value of the difference between the same type of vital sign data at each time point and the adjacent previous time point to the time interval between adjacent time points is used as the feature value of the same type of vital sign data at each time point;

[0014] Each type of vital sign data is treated as a dimension of the data space to construct a multidimensional data space; a unique data point is determined in the multidimensional data space by using the feature values ​​of all types of vital sign data at each time point.

[0015] Preferably, determining the probability of abnormality for each vital sign data at each time point includes:

[0016] A preset number of adjacent moments before each moment are taken as adjacent historical moments, and the curve fitting results of each vital sign data at each moment and adjacent historical moments over time are taken as the fitting curve of each vital sign data at each moment.

[0017] The product of the variance of the fitting curve at each time point and the variance of the fitting deviation in adjacent historical data, and the sum of the fitting deviations, is used as the probability of abnormality for each vital sign at each time point.

[0018] Preferably, determining the probability of an anomaly for each data point at each time point includes:

[0019] The STL method of time series decomposition was used to decompose the fitted curve of each type of vital sign data at each time point, and the residual term and trend term were obtained respectively.

[0020] Calculate the difference distance between the trend terms and the difference distance between the residual terms of the fitted curves for any two different vital signs at each time point, and perform negative mapping on the difference distance between the residual terms;

[0021] The product of the difference distance between the trend terms and the negative mapping result is calculated and accumulated over all vital signs data at each time point as the monitoring anomaly degree at each time point;

[0022] The anomaly probability of a data point at each time point in the multidimensional data space is determined by using the monitoring anomaly degree at each time point and the mean of the anomaly probability of all types of vital signs data at each time point; wherein, the anomaly probability is negatively correlated with the monitoring anomaly degree and positively correlated with the mean.

[0023] Preferably, the method for obtaining the feature data points is as follows: using the global threshold segmentation method to obtain the threshold of the abnormal probability of all data points in the multidimensional data space, and taking the data points with an abnormal probability greater than the threshold as feature data points in the multidimensional data space.

[0024] Preferably, the step of dividing abnormal clusters and normal clusters includes:

[0025] The distribution anomaly index of each cluster is determined based on the differences in the distribution concentration of feature data points between each cluster and the other clusters, as well as the distance between the cluster centers.

[0026] Calculate the variance of the distribution of feature data points in each dimension of the multidimensional data space, and multiply the sum of the variance in all dimensions of the multidimensional data space with the distribution anomaly index as the difference index of each cluster.

[0027] A global threshold segmentation method is used to obtain the segmentation threshold of the difference index of all clusters. Clusters with difference index greater than the segmentation threshold are classified as abnormal clusters, and clusters with difference index less than or equal to the segmentation threshold are classified as normal clusters.

[0028] Preferably, determining the distribution anomaly index for each cluster includes:

[0029] Calculate the time interval between any two feature data points within each cluster and the range between feature data points within each cluster. Use the sum of the variance of the distribution of all the time intervals corresponding to each cluster and the range of feature data points within each cluster as the intra-cluster difference of each cluster.

[0030] Calculate the sum of Euclidean distances between each cluster and the cluster centers of all other clusters in the multidimensional data space. Use the product of the normalized sum of Euclidean distances and the intra-cluster differences as the distribution anomaly index for each cluster.

[0031] Preferably, the determination of the corrected vital sign data at the corresponding time point for each feature data point within the abnormal cluster includes:

[0032] The difference between the mean values ​​of each normal cluster and each abnormal cluster in the same dimension of the multidimensional data space is calculated as the numerator, and the ratio of the numerator to the mean value of each abnormal cluster in the same dimension of the multidimensional data space is used as the first difference.

[0033] The absolute value of the difference between the distribution variance of each normal cluster and each abnormal cluster in the same dimension of the multidimensional data space is used as the second difference value, and the ratio of the absolute value of the difference value to the distribution variance of each abnormal cluster in the same dimension of the multidimensional data space is used as the second difference value.

[0034] The product of the first difference and the second difference is used as the correction coefficient of each normal cluster for each abnormal cluster in one dimension;

[0035] The mean of the correction coefficients of all normal clusters to each abnormal cluster across all dimensions is taken as the correction degree of each abnormal cluster.

[0036] Preferably, the real-time monitoring results of elderly users' vital signs data based on the correction results include:

[0037] For each feature data point within each abnormal cluster, the product of each vital sign data collected at the corresponding time for each feature point and the correction degree of each abnormal cluster is taken as the correction amount of each vital sign data at the corresponding time. The sum of the correction amount of each vital sign data and each vital sign data collected at the corresponding time is taken as the correction result of each vital sign data at the corresponding time.

[0038] Preferably, the real-time monitoring results of elderly users' vital signs data based on the correction results include:

[0039] Each type of vital sign data collected at the corresponding time point for each feature data point within the abnormal cluster is corrected;

[0040] For any type of vital sign data, arrange all the correction results and the normally collected data for each type of vital sign data in chronological order to obtain the monitoring results of each type of vital sign data for elderly users within the monitoring period.

[0041] This application has at least the following beneficial effects:

[0042] This application addresses the issue that existing wearable devices primarily monitor human vital signs by sampling the data. While this method is convenient and widely used, wearable devices may move during human activity, leading to discrepancies between the collected vital sign data and the actual changes. Therefore, this application first uses vital sign data collected at adjacent time points to determine the data point corresponding to each time point in a multidimensional data space. It then performs a preliminary data space mapping on the vital sign data collected at each time point based on the instantaneous changes in the data, facilitating subsequent analysis based on data changes rather than the collected data itself, thus avoiding the influence of isolated noise points. Secondly, it determines the anomaly probability of each type of vital sign data at each time point based on the fitted curves of the vital sign data collected at each time point and historical time points. Based on the anomaly probability and the difference between the trend terms of the fitted curves of different types of vital sign data at each time point, it determines the anomaly probability of the corresponding data point at each time point. Finally, it analyzes the data changes caused by wearable device movement and abnormalities in the elderly user's own behavior. The system assesses the probability of abnormalities in the collected vital sign data at each time point based on the different variations. Then, it clusters the feature data points selected based on the abnormality probability, and divides them into abnormal clusters and normal clusters based on the aggregation characteristics and intra-cluster distribution of feature data points within different clusters. This clustering of feature data points facilitates subsequent adaptive correction of the vital sign data at the corresponding time point for feature data points within abnormal clusters by comparing the distribution differences between normal and abnormal clusters. Finally, based on the values ​​of feature data points within abnormal and normal clusters, the correction result for the vital sign data at the corresponding time point for each feature data point within the abnormal cluster is determined, and the real-time monitoring results of the elderly user's vital sign data are obtained based on the correction results. This application improves the accuracy of vital sign data monitoring results by analyzing the overall and local variation characteristics of the collected data and adaptively correcting abnormal vital sign data. Attached Figure Description

[0043] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating the real-time monitoring method for elderly vital signs based on wearable devices provided in this application embodiment. Detailed Implementation

[0045] To further illustrate the technical means and effects adopted by this application in order to achieve the intended purpose of the invention, the following detailed description of the method for real-time monitoring of elderly vital signs based on wearable devices proposed in this application is provided in conjunction with the accompanying drawings and preferred embodiments.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0047] The following description, in conjunction with the accompanying drawings, details the specific scheme of the real-time monitoring method for elderly vital signs based on wearable devices provided in this application.

[0048] Example of a method for real-time monitoring of vital signs in the elderly based on wearable devices:

[0049] The specific scenario addressed in this embodiment is as follows: When using wearable devices to monitor the health of elderly users, the wearable device is first worn by the elderly user. During the process of collecting vital sign data from the elderly user through its built-in smart sensors, the accuracy of the collected data may be affected by factors such as device displacement and sweat during the elderly user's movement, leading to deviations in the vital sign monitoring results. Therefore, it is necessary to perform overall and local feature analysis on the collected data and correct any abnormal data to a certain extent, thereby improving the accuracy of the vital sign monitoring results. The wearable device includes smartwatches, smart bracelets, smart vests, etc.

[0050] Please see Figure 1 The document presents a flowchart of a method for real-time monitoring of vital signs in the elderly based on wearable devices, according to an embodiment of this application. The method includes the following steps:

[0051] Step S1: Collect vital sign monitoring data of elderly users using wearable devices.

[0052] This application aims to monitor the vital signs of elderly users through wearable devices. First, the wearable device is worn by the subject to collect vital sign data, including but not limited to heart rate, blood pressure, body temperature, respiratory rate, and step count. In this embodiment, the preset collection duration is 10 minutes, and the data collection frequency is 60Hz per minute, meaning that vital sign data is collected once per second. In other embodiments, the implementer can preset the collection duration and frequency according to the elderly user's vital signs; this application does not impose any special restrictions on this.

[0053] Step S2: Analyze the difference between the changes in vital signs data caused by the movement of wearable devices and the changes in vital signs data caused by abnormal conditions of elderly users, calculate the probability of anomalies in the data points in the constructed multidimensional data space, and then determine the feature data points in the multidimensional data space.

[0054] When elderly users wear wearable devices during exercise, or when sweat accumulates between the wearable device and their skin, the accuracy of vital sign data collection can be affected. This embodiment uses heart rate as an example to analyze the errors in the heart rate data collection process and corrects abnormal heart rate data to obtain accurate heart rate data.

[0055] Specifically, while the movement of wearable devices may appear slight to a human, subtle changes in skin texture and other minute details can alter the path of light from the sensor, leading to errors in the collected data. This is because heart rate monitoring primarily involves collecting data on changes in blood volume and hemoglobin absorption to indirectly calculate the heart rate.

[0056] Furthermore, let's take the i-th moment as an example for any given moment in the data acquisition process. First, the ratio of the absolute value of the difference between the number of steps collected at the i-th moment and the number of steps collected at the (i-1)-th moment to the sampling interval is used as the velocity feature value at the i-th moment. Similarly, for the other types of vital sign data collected by the wearable device, the ratio of the absolute value of the difference between the data collected at the i-th moment and the number of steps collected at the (i-1)-th moment to the sampling interval is calculated as the feature value of the vital sign data. Second, after obtaining the feature values ​​of all types of vital sign data at each moment, the feature values ​​of each type of vital sign data are mapped to a dimension to construct a multi-dimensional data space. The feature values ​​of all types of vital sign data at each moment determine a unique data point in the multi-dimensional data space.

[0057] Under normal circumstances, the heart's contraction and relaxation exhibit a clear periodicity, and heart rate fluctuates within a certain range. When elderly users are exercising, their heart rate, body temperature, and respiratory rate also change significantly with the intensity of exercise. Furthermore, heart rate, body temperature, and respiratory rate all increase with the intensity of exercise and decrease with the slowing of exercise, meaning that different types of vital signs show relatively similar trends over the same time period. However, changes in the distance between the wearable device and the elderly user's skin during exercise, or the generation of sweat at the wearing site, can affect these different types of vital signs to varying degrees, disrupting the aforementioned similar trends.

[0058] Specifically, the k previous times for each time point are taken as adjacent historical times, and the data collected for each vital sign at each time point are subjected to curve fitting in time. The fitting results are used as the fitting curves for each vital sign data at each time point.

[0059] Secondly, after obtaining the fitted curves for all vital signs at each time point, the STL (Standard Time Series) method is used to decompose the fitted curves for each type of vital sign at each time point, obtaining the residual term, trend term, and seasonal term. The STL method is a well-known technique, and its specific process will not be elaborated further.

[0060] Then, the difference distances between the trend terms and the residual terms of the fitted curves for different vital signs at time i are calculated respectively. The difference distance is used to measure the degree of difference between the trend terms and the residual terms. Differential distances that can be used in different embodiments include, but are not limited to, Euclidean distance, DTW distance, and value variance.

[0061] It should be noted that curve fitting is a common technique in the field of data processing. The specific process will not be elaborated here. Common curve fitting methods include, but are not limited to, least squares fitting and polynomial fitting. This application does not impose any special limitations on the specific method of curve fitting. Preferably, in one embodiment of this application, least squares fitting is used to obtain the fitting curves for each vital sign data at k time points.

[0062] It should be further explained that obtaining data from k historical moments at each moment is to analyze the data change trend at each moment and to avoid the influence of noise points on the data at a single moment, thereby improving the accuracy of subsequent data analysis. Under the premise of achieving the above objectives, the value of k can be chosen differently in different embodiments, and this application does not impose any special restrictions on it. Preferably, in one embodiment of this application, the value of k is set to 7. For moments with fewer than k historical moments, data from k historical moments is obtained using data padding methods. Data padding includes, but is not limited to, nearest neighbor padding, mean padding, interpolation padding, etc. Data padding is a well-known technique, and the specific process will not be described in detail.

[0063] Furthermore, when using wearable devices to monitor the vital signs of elderly users, they are usually guided to perform some health-preserving exercises, such as slow walking or calisthenics, to more accurately reflect their health status. Changes in the distance between the wearable device and the elderly user's skin during exercise, or sweating at the wearing site, can affect different types of vital sign data. Since elderly users typically have relatively long periods of rest during exercise, this should affect the vital sign data at several consecutive collection points. Moreover, the degree of influence on the vital sign data will change depending on the intensity of the exercise or the duration of the exercise.

[0064] First, obtain the fitted curve at time i. Calculate the fitting deviation for each of the (k+1) time points. A larger difference in fitting deviation among the (k+1) time points indicates a more significant change in the instantaneous trend of the vital signs data at different times. Conversely, a smaller fitting deviation among the (k+1) time points, and a smaller difference in fitting deviation among the (k+1) time points, indicates a more stable trend in the vital signs data among the (k+1) time points. Here, the similarity of the trend characteristics between the fitted curves of different vital signs at the same time point is used to preliminarily assess the probability of abnormality of the vital signs data at each time point, thus obtaining the probability of data abnormality at each time point. Calculate the probability of abnormality of the j-th vital sign data at time i:

[0065] t i,j =σ i,j ×z i,j

[0066] In the formula, t i,j σ is the probability of anomalies in the j-th vital sign data at time i. i,j z is the variance of the fitting deviation at time k+1 on the fitted curve of the j-th vital sign data at time i. i,j It is the sum of the fitting deviations at k+1 times on the fitted curve of the j-th vital sign data at time i.

[0067] Secondly, by combining the stability of the trend characteristics of the fitted curves of various vital signs at adjacent time points, and the magnitude of the time interval between the corresponding data point and its adjacent data points in the multidimensional data space, the probability of anomalies for each data point is further evaluated. First, for the i-th time point, the difference distance d between the trend terms of the fitted curves of any two different vital signs at the i-th time point is calculated. i,1 The difference distance d between residual terms i,2 , will d i,2 The negative mapping result is used as the weight pair for d i,1 A weighted average is applied, and the sum of the weighted results across all vital sign data at time i is used as the degree of monitoring anomaly at time i. This calculation is based on the fact that a smaller residual indicates less interference from noise and other factors during data acquisition, resulting in higher data accuracy; and a smaller distance between residuals indicates a more consistent degree of interference between the two types of vital sign data at time i.

[0068] Then, the anomaly probability of data point p in the multidimensional data space is determined by using the monitoring anomaly degree at time i and the mean of the anomaly probabilities of all types of vital signs data at time i. The anomaly probability is negatively correlated with the monitoring anomaly degree and positively correlated with the mean.

[0069] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship is determined by the actual application, and this application does not impose any special restrictions.

[0070] It should be noted that negative correlation means that when one variable increases, the other variable decreases accordingly, and the two variables change in opposite directions. When one variable changes from large to small or from small to large, the other variable also changes from small to large or from large to small. The specific relationship is determined by practical application, and this application does not impose any special restrictions.

[0071] Preferably, in one embodiment of this application, the formula for calculating the anomaly probability of data point p determined in the multidimensional data space at the i-th time moment is as follows:

[0072]

[0073] In the formula, r i This represents the degree of abnormality detected at time i, M is the type of vital sign data, j and n represent the j-th and n-th types of vital sign detection data, respectively, and d i,1 (j,n),d i,2 (j,n) are the difference distance between the trend terms and the difference distance between the residual terms of the fitted curves of the j-th and n-th vital sign detection data at the i-th time, respectively. μ is a constant parameter used to prevent the denominator from being 0. Here, it can be taken as 0.001. The specific value can be adjusted according to the implementation scenario. This application does not impose any special restrictions.

[0074] u p It is the probability of an anomaly in data point p. It is the mean of the abnormal probabilities of all types of vital signs at time i.

[0075] Furthermore, based on the above method, the anomaly probability of each data point in the multidimensional data space can be obtained. The higher the anomaly probability, the higher the likelihood of the data point being abnormal. A global threshold segmentation method is used to obtain a threshold for all the anomaly probabilities, and data points with anomaly probabilities greater than the threshold are taken as feature data points. Threshold segmentation is a commonly used technique in the field of data processing; the specific process will not be elaborated further.

[0076] Step S3: Cluster the feature data points in the multidimensional data space and divide them into abnormal clusters and normal clusters based on the clustering characteristics and distribution of feature data points within different clusters.

[0077] Specifically, if the problem at time i is caused by a sudden abnormality in the elderly user's health condition during the vital sign monitoring process, then the various characteristic data collected by the sensors in the wearable device at time i will show significant changes. Furthermore, the frequency of sudden abnormalities in the elderly user's health condition throughout the entire vital sign monitoring cycle should be very low, and the corresponding data points in the multidimensional data space will not be concentrated. However, if the problem stems from changes in the distance between the wearable device and the elderly user's skin during exercise, or the impact of sweat on the sensor's collection of different vital sign data due to the intermittent nature of exercise, the corresponding data points in the multidimensional data space will exhibit a certain degree of concentration. For example, if an elderly user takes a 5-minute slow walk after a meal, the data points corresponding to the vital sign data changes caused by exercise within a few minutes after each of the three daily meals will show a clear clustering.

[0078] Furthermore, after obtaining the feature data points in the multidimensional data space according to the above steps, a clustering algorithm is used to divide all feature data points into several clusters.

[0079] It should be noted that data clustering is a common technique in the field of data processing, and the specific process will not be elaborated further. Commonly used data clustering algorithms include, but are not limited to, K-means clustering, AP (Affinity Propagation) clustering, and hierarchical clustering. This application does not impose any special restrictions on the clustering method for feature data points. Preferably, this application uses the AP clustering algorithm to obtain the clustering results of the feature data points.

[0080] Secondly, for any given cluster, let's take the a-th cluster as an example for further analysis. If the intervals between the time points corresponding to the feature data points within the a-th cluster are roughly similar and exhibit a relatively stable periodicity, then the feature data points within the a-th cluster are more likely to correspond to different collection times during the elderly user's regular daily exercise. If the intervals between the time points corresponding to the feature data points within the a-th cluster vary significantly, and the data points within the a-th cluster exhibit strong dispersion, then the feature data points within the a-th cluster are more likely to correspond to moments when the elderly user's health condition suddenly deteriorates during vital sign monitoring.

[0081] Here, a difference index is calculated for each cluster to reflect the difference in the distribution of feature data points between each cluster and the other clusters. The formula for calculating the difference index of the a-th cluster is:

[0082] L a =norm(d a )×(σ(T a )+C a )

[0083] In the formula, L aIt is the distribution anomaly index of the a-th cluster, d a It is the sum of the Euclidean distances between the cluster centers of the a-th cluster and the cluster centers of the other clusters in the multidimensional data space, where norm() is the normalization function, and T a It is the set of time intervals between the corresponding collection times of feature data points within the a-th cluster, σ(T) a ) is a set T a The variance of the time interval, C a It is the range between feature data points within the a-th cluster;

[0084] Secondly, the difference index of the a-th cluster is determined by combining the differences in the values ​​of the feature data points among the a-th clusters in each dimension of the multidimensional data space:

[0085]

[0086] In the formula, V a σ is the difference index of the a-th cluster, M is the dimension of the multidimensional data space, and its size is equal to the number of types of vital sign data collected. σ is the m-th dimension of the multidimensional data space. c,m It is the variance of the distribution of the feature data points in the a-th cluster along the m-th dimension.

[0087] Among them, (σ(T) a )+C a The data reflects the intra-cluster differences of the a-th cluster. These intra-cluster differences characterize the periodicity of the time intervals and the discreteness of the positions of the feature data points within the a-th cluster. The distance d between the a-th cluster and the other clusters is then considered. a Normalization is performed, and the intra-class differences are weighted to obtain the distribution anomaly index L. a The larger the value, the more irregular the distribution of feature data points within the cluster; and the value of a feature data point on the same dimension is the feature value of the next type of vital sign data at the corresponding acquisition time, therefore the distribution variance σ c,m The larger the value, the greater the difference in vital sign data changes among data points in the a-th cluster at adjacent time points.

[0088] Furthermore, the difference index of each cluster is obtained separately, and a segmentation threshold for the difference index of all clusters is obtained using a global threshold segmentation method. Clusters with difference indices greater than the segmentation threshold are classified as abnormal clusters affected by the movement of wearable devices, while clusters with difference indices less than or equal to the segmentation threshold are classified as normal clusters not affected by the movement of wearable devices. Global threshold segmentation is a commonly used technique in the field of data processing, and its specific process will not be elaborated upon here.

[0089] S4. Based on the value of feature data points in abnormal clusters and normal clusters, determine the correction result of vital sign data at the corresponding time for each feature data point in the abnormal cluster, thereby obtaining the real-time monitoring result of vital sign data of elderly users.

[0090] In step S3 of this embodiment, abnormal clusters are identified. The heart rate corresponding to the feature data points in the abnormal clusters at the time of collection is more likely to be abnormal data caused by changes due to the movement of the wearable device. Therefore, this embodiment will correct the heart rate corresponding to the feature data points in the abnormal clusters at the time of collection. For abnormal clusters, the greater the deviation compared to the normal vital signs data collected at the corresponding time of the feature data points in normal clusters, the greater the degree of correction required.

[0091] Here, the correction degree of each abnormal cluster is calculated to characterize the degree of correction required compared to the vital signs data collected at the corresponding time points of the feature data points within the normal clusters.

[0092] Specifically, for the b-th anomalous cluster, the formula for calculating the correction degree corresponding to the b-th anomalous cluster is as follows:

[0093]

[0094] In the formula, q b,a (m) is the correction coefficient of the a-th normal cluster on the m-th abnormal cluster for the b-th abnormal cluster. These are the mean and σ values ​​of all feature data points in the a-th normal cluster and the b-th abnormal cluster, respectively, along the m-th dimension. a,m σ b,m These are the variances of the distributions of all feature data points in the m-th dimension within the a-th normal cluster and the b-th abnormal cluster, respectively.

[0095] Q b is the correction degree of the b-th abnormal cluster, and N is the number of normal clusters.

[0096] It should be noted that the variance of the distribution reflects the degree of data fluctuation, which is obtained through the second difference. The distribution deviation of feature data points in the m-th dimension within the b-th abnormal cluster compared to the a-th normal cluster is represented by the first difference. It reflects the numerical deviation of the feature data points in the m-th dimension within the two clusters. The larger the distribution deviation and data deviation, the more severe the influence of the wearable device's movement on the feature data points in the b-th abnormal cluster at the time of collection, and the higher the degree of correction.

[0097] Furthermore, according to the above process, the correction degree of each abnormal cluster is obtained, and the collected heart rate data is corrected based on the correction degree.

[0098] Specifically, for each feature data point in the b-th abnormal cluster, the product of each vital sign data collected at the corresponding time of each feature point and the correction degree of the b-th abnormal cluster is used as the correction amount of each vital sign data at the corresponding time, and the sum of the correction amount of each vital sign data and each vital sign data collected at the corresponding time is used as the correction result of each vital sign data at the corresponding time.

[0099] Next, for any feature data point within each abnormal cluster, the vital sign data collected at the corresponding time for each feature data point within the abnormal cluster are corrected according to the steps described above, thus obtaining the corrected results for the vital sign data collected at the corresponding time for all feature data points within the abnormal clusters. Finally, for any type of vital sign data, all corrected results for each type of vital sign data are inserted into the normally collected data in chronological order to obtain the monitoring results for each type of vital sign data for elderly users within the monitoring period.

[0100] It should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0101] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for real-time monitoring of vital signs in the elderly based on wearable devices, characterized in that, The method includes the following steps: Collect vital sign data of elderly users using wearable devices; The method involves using vital sign data collected at adjacent time points to determine the data point corresponding to each time point in a multidimensional data space. This includes: using the ratio of the absolute value of the difference between the same vital sign data at each time point and the adjacent time point to the time interval between adjacent time points as the feature value of the same vital sign data at each time point; constructing a multidimensional data space by treating each type of vital sign data as one dimension of the data space; and determining a unique data point in the multidimensional data space using the feature values ​​of all types of vital sign data at each time point. The abnormal probability of each vital sign data at each time point is determined based on the fitted curves of the vital sign data collected at each time point and historical time points. This includes: taking a predetermined number of adjacent time points before each time point as adjacent historical time points; taking the curve fitting results of each vital sign data at each time point and adjacent historical time points as the fitted curves of each vital sign data at each time point; taking the product of the distribution variance of the fitting deviation of the fitted curve at each time point and adjacent historical time points and the sum of the fitting deviations as the abnormal probability of each vital sign data at each time point; and determining the abnormal probability of the corresponding data point at each time point based on the abnormal probability and the difference between the trend terms of the fitted curves of different vital sign data at each time point, including: using the Time Series Decomposition (STL) method to separately decompose the data into different time series. The fitted curves for each type of vital sign data at each time point are decomposed to obtain residual terms and trend terms. The difference distance between the trend terms and the difference distance between the residual terms of the fitted curves for any two different types of vital sign data at each time point are calculated, and a negative mapping is applied to the difference distance between the residual terms. The product of the difference distance between the trend terms and the negative mapping result is calculated and summed across all types of vital sign data at each time point to determine the monitoring anomaly level at each time point. The anomaly level at each time point and the mean of the anomaly probabilities of all types of vital sign data at each time point are used to determine the anomaly probability of the data points identified at each time point in the multidimensional data space. The anomaly probability is negatively correlated with the monitoring anomaly level, and positively correlated with the mean. Clustering is performed on the feature data points used for anomaly detection. Based on the clustering characteristics and intra-cluster distribution of feature data points within different clusters, abnormal clusters and normal clusters are identified. This includes: calculating the time interval between any two feature data points within each cluster at corresponding acquisition times; using the sum of the variance of all time intervals for each cluster and the range between feature data points within each cluster as the intra-cluster variance of each cluster; calculating the cumulative Euclidean distance between the cluster centers of each cluster and all other clusters in the multidimensional data space; and converting the Euclidean distance... The product of the normalized sum of distances and the intra-cluster variance is used as the distribution anomaly index for each cluster. The variance of the feature data points within each cluster in each dimension of the multidimensional data space is calculated. The product of the sum of the variances in all dimensions of the multidimensional data space and the distribution anomaly index is used as the variance index for each cluster. A global threshold segmentation method is used to obtain the segmentation threshold for the variance index of all clusters. Clusters with variance indices greater than the segmentation threshold are classified as anomalous clusters. Clusters with variance indices less than or equal to the segmentation threshold are classified as normal clusters. Based on the values ​​of feature data points within abnormal and normal clusters, the correction results of vital sign data for each feature data point within the abnormal cluster at the corresponding time are determined, and the real-time monitoring results of vital sign data for elderly users are obtained based on the correction results.

2. The method for real-time monitoring of vital signs in the elderly based on wearable devices according to claim 1, characterized in that, The method for obtaining feature data points is as follows: the global threshold segmentation method is used to obtain the threshold of the anomaly probability of all data points in the multidimensional data space, and the data points with an anomaly probability greater than the threshold are used as feature data points in the multidimensional data space.

3. The method for real-time monitoring of vital signs in the elderly based on wearable devices according to claim 1, characterized in that, Determine the correction results of vital sign data for each feature data point within the abnormal cluster at the corresponding time point, including: The difference between the mean values ​​of each normal cluster and each abnormal cluster in the same dimension of the multidimensional data space is calculated as the numerator, and the ratio of the numerator to the mean value of each abnormal cluster in the same dimension of the multidimensional data space is used as the first difference. The absolute value of the difference between the distribution variance of each normal cluster and each abnormal cluster in the same dimension of the multidimensional data space is used as the second difference value, and the ratio of the absolute value of the difference value to the distribution variance of each abnormal cluster in the same dimension of the multidimensional data space is used as the second difference value. The product of the first difference and the second difference is used as the correction coefficient of each normal cluster for each abnormal cluster in one dimension; The mean of the correction coefficients of all normal clusters to each abnormal cluster across all dimensions is taken as the correction degree of each abnormal cluster.

4. The method for real-time monitoring of vital signs in the elderly based on wearable devices according to claim 1, characterized in that, Based on the correction results, real-time monitoring results of vital signs data for elderly users were obtained, including: For each feature data point within each abnormal cluster, the product of each vital sign data collected at the corresponding time and the correction degree of each abnormal cluster is taken as the correction amount of each vital sign data at the corresponding time. The sum of the correction amount of each vital sign data and each vital sign data collected at the corresponding time is taken as the correction result of each vital sign data at the corresponding time.

5. The method for real-time monitoring of vital signs in the elderly based on wearable devices according to claim 1, characterized in that, Based on the correction results, real-time monitoring results of vital signs data for elderly users were obtained, including: Each type of vital sign data collected at the corresponding time point for each feature data point within the abnormal cluster is corrected; For any type of vital sign data, arrange all the correction results and the normally collected data for each type of vital sign data in chronological order to obtain the monitoring results of each type of vital sign data for elderly users within the monitoring period.

Images