Human body health index condition monitoring and analyzing system and method

Abstract

The invention relates to the technical field of data acquisition and signal transmission, and discloses a human health index condition monitoring and analyzing system and method, and the method comprises the steps: configuring a sensor network of a target user, collecting the multi-sensing data of the target user in real time, extracting the feature vector of the multi-sensing data, and transmitting the feature vector to the target user; performing data fusion on the multi-sensing data, defining an anomaly detection algorithm of the fused sensing data, calculating an outlier coefficient of the window data, and identifying abnormal data in the abnormal data to be judged; the method comprises the following steps: acquiring abnormal data of a target user, performing abnormal restoration on the abnormal data, defining a weak signal extraction algorithm of the restoration data, extracting a bio-electricity signal of the restoration data, reconstructing the bio-electricity signal, extracting a human body health parameter of the reconstructed bio-electricity signal, and analyzing a health index of the target user. According to the invention, the comprehensiveness and accuracy of human health index condition monitoring can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a human health indicator status monitoring and analysis system and method. Background Art

[0002] Monitoring and analysis of human health indicators refers to the process of monitoring, recording, evaluating and analyzing the health status of individuals through a series of methods and tools. Monitoring and analysis of human health indicators has become an important part of the public health system, aiming to improve the health level of the whole people and build a harmonious society.

[0003] At present, the existing methods for monitoring and analyzing human health indicators include subjective survey method and physiological parameter detection method. These methods mainly collect and monitor single data in real time, and use a single data source to assess human health. A single indicator cannot fully reflect the health status of an individual, leading to misunderstanding or neglect of other potential health problems. Summary of the invention

[0004] The present invention provides a human health index status monitoring and analysis system and method, the main purpose of which is to improve the comprehensiveness and accuracy of human health index status monitoring.

[0005] To achieve the above-mentioned purpose, the present invention provides a human health index status monitoring and analysis system, comprising: a fusion sensor data module, an abnormality detection module, an abnormal data analysis module, a bioelectric signal extraction module and a health index analysis module;

[0006] The fusion sensor data module is used to configure a sensor network of a target user, collect multi-sensor data of the target user in real time based on the sensor network, extract feature vectors of the multi-sensor data, analyze the multi-sensor data based on the feature vectors to obtain multi-analysis results, and fuse the multi-sensor data based on the multi-analysis results to obtain fused sensor data;

[0007] The anomaly detection module is used to define an anomaly detection algorithm for the fused sensor data, extract window data of the fused sensor data using a trained sliding window model, detect an outlier factor of the window data according to the anomaly detection algorithm, calculate an outlier coefficient of the window data based on the outlier factor, and when the outlier coefficient is greater than a preset threshold, classify the window data corresponding to the outlier factor as abnormal data to be determined;

[0008] The abnormal data analysis module is used to calculate the single-mode abnormal probability of the abnormal data to be judged, determine the multi-mode abnormal probability of the abnormal data to be judged according to the single-mode abnormal probability, and identify the abnormal data in the abnormal data to be judged based on the multi-mode abnormal probability;

[0009] The bioelectric signal extraction module is used to perform abnormal repair on the abnormal data to obtain repair data, define a weak signal extraction algorithm for the repair data, extract the bioelectric signal of the repair data based on the weak signal extraction algorithm, decompose the bioelectric signal to obtain a low-frequency component and a high-frequency component, and reconstruct the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal;

[0010] The health index analysis module is used to extract human health parameters from the reconstructed bioelectric signal, and analyze the health index of the target user based on the human health parameters.

[0011] A method for monitoring and analyzing human health indicators, characterized in that the method comprises:

[0012] configuring a sensor network of a target user, collecting multi-sensor data of the target user in real time based on the sensor network, extracting feature vectors of the multi-sensor data, analyzing the multi-sensor data based on the feature vectors to obtain multi-analysis results, and fusing the multi-sensor data based on the multi-analysis results to obtain fused sensor data;

[0013] defining an anomaly detection algorithm for the fused sensor data, extracting window data of the fused sensor data using a trained sliding window model, detecting an outlier factor of the window data according to the anomaly detection algorithm, calculating an outlier coefficient of the window data based on the outlier factor, and classifying the window data corresponding to the outlier factor as abnormal data to be determined when the outlier coefficient is greater than a preset threshold;

[0014] Calculating the single-mode abnormality probability of the abnormal data to be determined, determining the multi-mode abnormality probability of the abnormal data to be determined according to the single-mode abnormality probability, and identifying the abnormal data in the abnormal data to be determined based on the multi-mode abnormality probability;

[0015] Performing abnormal repair on the abnormal data to obtain repaired data, defining a weak signal extraction algorithm for the repaired data, extracting a bioelectric signal of the repaired data based on the weak signal extraction algorithm, decomposing the bioelectric signal to obtain a low-frequency component and a high-frequency component, and reconstructing the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal;

[0016] Extract human health parameters of the reconstructed bioelectric signal, and analyze the health index of the target user based on the human health parameters.

[0017] The embodiment of the present invention can comprehensively monitor multiple health indicators of the human body by configuring the sensor network of the target user, and provide a comprehensive health status assessment; optionally, the embodiment of the present invention can reduce the error or uncertainty that may exist in a single sensor by fusing the multi-sensor data based on the multi-analysis results, thereby improving the accuracy of the overall data; the embodiment of the present invention can ensure the stability and security of the system while improving the detection performance and more accurately identifying the abnormal mode by defining the abnormal detection algorithm of the fused sensor data; the embodiment of the present invention can more comprehensively evaluate the abnormal degree of the data by determining the multi-mode abnormal probability of the abnormal data to be judged according to the single-mode abnormal probability, thereby improving the accuracy of abnormal detection; the embodiment of the present invention can extract the bioelectric signal of the repair data based on the weak signal extraction algorithm, which can effectively suppress noise and improve the signal-to-noise ratio of the bioelectric signal, making the signal clearer and facilitating subsequent analysis and processing. Finally, the embodiment of the present invention can track the health status of the individual in real time by extracting the human health parameters of the reconstructed bioelectric signal, thereby improving the accuracy of health analysis. Therefore, the present invention can improve the comprehensiveness and accuracy of human health indicator status monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A functional module diagram of a human health indicator status monitoring and analysis system provided by an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of a flow chart of a method for monitoring and analyzing a human health indicator status provided by an embodiment of the present invention;

[0020] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In addition, the step sequence in the following method embodiments is only an example and not a strict limitation.

[0023] In fact, the server-side device deployed by the human health index status monitoring and analysis system may be composed of one or more devices. The above-mentioned human health index status monitoring and analysis system can be implemented as: business instance, virtual machine, hardware device. For example, the human health index status monitoring and analysis system can be implemented as a business instance deployed on one or more devices in the cloud node. In simple terms, the live broadcast service system can be understood as a software deployed on the cloud node, which is used to provide human health index status monitoring and analysis services to each user terminal. Alternatively, the human health index status monitoring and analysis system can also be implemented as a virtual machine deployed on one or more devices in the cloud node. Application software for managing each user terminal is installed in the virtual machine. Alternatively, the human health index status monitoring and analysis system can also be implemented as a server consisting of many hardware devices of the same or different types, and one or more hardware devices are set to provide human health index status monitoring and analysis services to each user terminal.

[0024] In terms of implementation, the human health indicator status monitoring and analysis system and the user end are adapted to each other. That is, the human health indicator status monitoring and analysis system is an application installed on the cloud service platform, and the user end is a client that establishes a communication connection with the application; or the human health indicator status monitoring and analysis system is implemented as a website, and the user end is implemented as a web page; or the human health indicator status monitoring and analysis system is implemented as a cloud service platform, and the user end is implemented as a small program in an instant messaging application.

[0025] Reference Figure 1 , which is a functional module diagram of a human health indicator status monitoring and analysis system provided by an embodiment of the present invention.

[0026] The human health index status monitoring and analysis system 100 of the present invention can be set in a cloud server. In terms of implementation, it can be used as one or more service devices, or it can be installed as an application on the cloud (such as a server, server cluster, etc. for human health index status monitoring and analysis), or it can be developed as a website. According to the functions implemented, the human health index status monitoring and analysis system 100 includes a fusion sensor data module 101, an anomaly detection module 102, an anomaly data analysis module 103, a bioelectric signal extraction module 104, and a health index analysis module 105.

[0027] In the embodiment of the present invention, in the tracking based on the monitoring and analysis of the human health index status, each of the above modules can be implemented independently and called with other modules. The call here can be understood as that a certain module can connect to multiple modules of another type and provide corresponding services to the multiple modules connected to it. In the human health index status monitoring and analysis system provided by the embodiment of the present invention, there is no need to modify the program code. The scope of application of the human health index status monitoring and analysis architecture can be adjusted by adding modules and directly calling them to achieve cluster-based horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the human health index status monitoring and analysis system. In actual applications, the above modules can be set in the same device or different devices, or they can be set in virtual devices, such as service instances in cloud servers.

[0028] In the following, in conjunction with specific embodiments, each component and specific workflow of the human health indicator status monitoring and analysis system are described respectively.

[0029] The fused sensor data module 101 is used to configure a sensor network of a target user, collect multi-sensor data of the target user in real time based on the sensor network, extract feature vectors of the multi-sensor data, analyze the multi-sensor data based on the feature vectors to obtain multi-analysis results, and fuse the multi-sensor data based on the multi-analysis results to obtain fused sensor data.

[0030] The embodiment of the present invention can comprehensively monitor multiple health indicators of the human body and provide a comprehensive health status assessment by configuring a sensor network of a target user. The sensor network refers to a system composed of multiple sensor nodes, which are interconnected by wire or wireless means and work together to monitor, collect, process and transmit data.

[0031] Optionally, as an embodiment of the present invention, the sensor network for configuring the target user may be configured by centralized assembly.

[0032] The embodiment of the present invention can monitor the key physiological parameters of the target user in real time by collecting the multi-sensor data of the target user in real time based on the sensor network, which is helpful for analyzing the health indicators of the target user. The multi-sensor data refers to a data set collected by multiple different types of sensors.

[0033] The embodiment of the present invention can achieve effective data fusion by extracting the feature vector of the multi-sensor data, thereby obtaining a more comprehensive data analysis result. The feature vector refers to a set of values ​​extracted from the original multi-sensor data, which can represent the essential attributes or key information of the original data.

[0034] As an embodiment of the present invention, the extracting the feature vector of the multi-sensor data includes:

[0035] Determining feature extraction requirements for the multi-sensor data, and performing data cleaning on the multi-sensor data to obtain cleaned data;

[0036] Analyzing the data types of the cleaned data, wherein the data types include: time series data, waveform data and image data;

[0037] Extracting data features of the cleaned data according to the data type;

[0038] Calculating the feature correlation between the data feature and the feature extraction requirement, and analyzing the feature importance of the data feature using a trained feature importance analysis model;

[0039] Based on the feature relevance and the feature importance, the data features are screened to obtain feature vectors.

[0040] Among them, the feature extraction requirements refer to the specific requirements and standards set for extracting features from multi-sensor data in order to achieve specific data analysis, model building or decision support goals. The cleaned data refers to the original data set that has been subjected to a series of data preprocessing operations to remove noise, outliers, redundant information and incomplete data points. The data type refers to the classification of data into different categories according to the characteristics and uses of the data. The time series data refers to a series of data points arranged in chronological order, which is usually used to record the changes of a certain variable over time, such as body temperature, heart rate, blood sugar, etc. The waveform data refers to data that is usually presented in the form of a continuous curve, such as electrocardiogram and electroencephalogram. The image data refers to visual information represented in two-dimensional or three-dimensional form, usually composed of pixel arrays, each pixel containing color and intensity information, such as X-ray images, magnetic resonance images, etc. The data feature refers to representative or distinguishing information extracted from the original data, which can be used to describe the attributes, laws or patterns of the data, such as statistical features, image features, frequency domain features, etc. The feature correlation refers to the relationship or degree of dependence between features and requirements. The feature importance analysis model refers to a model or method used to evaluate and quantify the influence of each feature in a data set on the prediction result. The feature importance refers to the influence of each feature in a data set on the prediction result or model output.

[0041] Optionally, extracting data features of the cleaned data according to the data type includes:

[0042] Calculate the mean, variance and standard deviation of the cleaned data according to the time series data corresponding to the data type;

[0043] Determining statistical characteristics of the cleaned data according to the mean, the variance, and the standard deviation;

[0044] Extracting waveform features of the cleaned data according to the waveform data corresponding to the data type;

[0045] According to the image data corresponding to the data type, using a preset image feature extraction technology to extract image features of the cleaned data;

[0046] The statistical features, the waveform features and the image features are integrated into data features.

[0047] Among them, the statistical features refer to parameters used to describe the distribution, central tendency and degree of dispersion of data. The waveform features refer to features extracted from the signal waveform that help describe or distinguish the signal, such as peak value, frequency component, frequency bandwidth, etc. The image feature extraction technology refers to a series of algorithms and methods used to identify and extract key information that helps image understanding and analysis from raw image data, such as color histogram, Canny edge detector, gray level co-occurrence matrix, etc. The image features refer to quantitative or qualitative information extracted from image data to describe image content, shape, color, texture or other attributes.

[0048] The embodiment of the present invention analyzes the multi-sensor data based on the feature vector to obtain multiple analysis results, which can serve as a bridge between different sensor data to achieve data fusion, thereby obtaining more comprehensive information. The multiple analysis results refer to multiple output results obtained by different analysis methods or models when processing and analyzing multi-sensor data.

[0049] Optionally, as an embodiment of the present invention, the multi-sensor data is analyzed based on the feature vector to obtain multiple analysis results, which can be obtained by a multi-modal data analysis method.

[0050] The embodiment of the present invention performs data fusion on the multi-sensor data based on the multi-analysis results to obtain fused sensor data, which can reduce the error or uncertainty that may exist in a single sensor, thereby improving the accuracy of the overall data. The fused sensor data refers to merging data sets from multiple sensors into a comprehensive data set through a specific data fusion technology.

[0051] As an embodiment of the present invention, the step of fusing the multi-sensor data based on the multi-analysis results to obtain fused sensor data includes:

[0052] Performing basic probability distribution on the multiple analysis results to obtain a probability distribution value;

[0053] Normalizing the multi-sensor data to obtain multi-normalized sensor data;

[0054] constructing a data fusion algorithm for the multi-normalized sensor data according to the probability distribution value;

[0055] Based on the data fusion algorithm, the multi-sensor data is fused to obtain initial fused data;

[0056] The confidence of the initial fusion data is calculated, and when the confidence meets a preset confidence interval, the initial fusion data is used as the fusion sensing data of the multi-sensing data.

[0057] Among them, the probability distribution value refers to the probability assigned to a set of events or hypotheses in probability theory or decision theory. The multi-normalized sensor data refers to a data set that is normalized by processing the data collected by multiple sensors so that the data of different sensors are on the same scale or range for easy comparison and fusion. The data fusion algorithm refers to an algorithm that combines data from multiple sources, sensors or information acquisition systems to obtain more accurate and comprehensive information. The initial fused data refers to the preliminary integrated data set obtained by combining the data or analysis results of multiple sensors by applying a specific data fusion algorithm in the early stage of the data fusion process. The confidence level refers to the assessment of the reliability or accuracy of a result or conclusion. The preset confidence interval refers to a confidence level that is pre-set according to needs or habits before starting statistical inference or data analysis.

[0058] Optionally, the data fusion algorithm for constructing the multi-normalized sensor data according to the probability distribution value can be constructed by utilizing a hierarchical fusion technology.

[0059] The anomaly detection module 102 is used to define an anomaly detection algorithm for the fused sensor data, use the trained sliding window model to extract the window data of the fused sensor data, detect the outlier factor of the window data according to the anomaly detection algorithm, and calculate the outlier coefficient of the window data based on the outlier factor. When the outlier coefficient is greater than a preset threshold, the window data corresponding to the outlier factor is classified as abnormal data to be determined.

[0060] The embodiment of the present invention can ensure system stability and security while improving detection performance and more accurately identifying abnormal patterns by defining the anomaly detection algorithm of the fused sensor data. The anomaly detection algorithm refers to a computational method or program for identifying anomalies or abnormal behaviors in a data set.

[0061] Optionally, as an embodiment of the present invention, the anomaly detection algorithm for defining the fused sensor data may be defined by a local anomaly factor method.

[0062] The embodiment of the present invention can effectively detect abnormal behaviors or events by extracting window data of the fused sensor data using a trained sliding window model, thereby improving the accuracy of abnormal detection. The sliding window model refers to a data processing technology that extracts a data subsequence by sliding a fixed-size window on a data sequence. The window data refers to subsequence data extracted from the original data sequence by the sliding window model.

[0063] The embodiment of the present invention can remove noise and erroneous data in the data set by detecting the outlier factor of the window data according to the anomaly detection algorithm, thereby improving the data quality and the reliability of the analysis results. The outlier factor refers to a data point that deviates from its neighboring points.

[0064] Optionally, as an embodiment of the present invention, the outlier factor of the window data detected according to the anomaly detection algorithm may be detected by a method of calculating a reachable distance using the anomaly detection algorithm.

[0065] The embodiment of the present invention can analyze the abnormality of each data point in the window data by calculating the outlier coefficient of the window data based on the outlier factor, so as to more accurately identify potential anomalies or outliers. Wherein, the outlier coefficient refers to an indicator used to measure the abnormality of each data point in the window data.

[0066] As an embodiment of the present invention, the calculating the outlier coefficient of the window data based on the outlier factor includes:

[0067] Identify the neighboring points of the outlier factor, and calculate the k-distance between the outlier factor and the neighboring points;

[0068] Analyzing the actual distance between the outlier factor and the neighboring point, and determining the reachable distance between the outlier factor and the neighboring point according to the actual distance and the k-distance;

[0069] Determining, according to the reachable distance, an outlier reachable density of the outlier factor and a neighboring point reachable density of the neighboring point;

[0070] According to the outlier reachable density and the neighboring point reachable density, the outlier coefficient of the window data is calculated using the following formula:

[0071]

[0072] Among them, θ(a) represents the outlier coefficient, a represents the outlier factor, b represents the neighboring point, and Ak (a) represents the set of k nearest neighbors of the outlier factor, M(b) represents the reachable density of neighboring points, and M(a) represents the reachable density of outliers.

[0073] Wherein, the neighboring points refer to the closest data points in the data set relative to the outlier factor. The k-distance refers to the distance from the outlier factor to the kth nearest neighbor point in the data set. The reachable distance refers to the maximum distance between the k-distances from a data point to its nearest neighbor point. The outlier reachable density refers to an indicator used to measure the degree of outlier of a data point relative to its neighborhood. The neighboring point reachable density refers to an indicator used to measure the density in the neighborhood of a point in a data set.

[0074] Optionally, the determining of the outlier factor and the reachable distance of the neighboring point based on the actual distance and the k-distance may be confirmed by comparing a maximum value of the actual distance and the k-distance.

[0075] In the embodiment of the present invention, when the outlier coefficient is greater than a preset threshold, the window data corresponding to the outlier factor is divided into abnormal data to be judged, so as to further analyze the cause of the data generation, thereby improving the data collection and processing process and improving the overall data quality. The preset threshold refers to a pre-defined critical value used to judge whether the outlier coefficient of the data point is high enough, and the threshold is usually set to 1.

[0076] The abnormal data analysis module 103 is used to calculate the single-mode abnormality probability of the abnormal data to be judged, determine the multi-mode abnormality probability of the abnormal data to be judged according to the single-mode abnormality probability, and identify the abnormal data in the abnormal data to be judged based on the multi-mode abnormality probability.

[0077] The embodiment of the present invention more accurately identifies data points that do not conform to the normal pattern by calculating the unimodal anomaly probability of the abnormal data to be determined, thereby reducing false positives and false negatives. The unimodal anomaly probability refers to the probability that a data point is considered unusual or abnormal in a specific data model or distribution.

[0078] As an embodiment of the present invention, the calculating the single-mode anomaly probability of the abnormal data to be determined includes:

[0079] Determine a probability distribution model of the abnormal data to be determined;

[0080] Extracting data parameters of the probability distribution model;

[0081] Calculate the probability density distribution function value of the abnormal data to be determined according to the data parameters;

[0082] The Z score of the abnormal data to be judged is calculated using the following formula:

[0083]

[0084] Among them, Z represents the Z score, Y represents the abnormal data to be judged, and P Y represents the mean of abnormal data, B Y Indicates abnormal data standards;

[0085] Determine the single-mode anomaly probability of the abnormal data to be determined based on the probability density distribution function value and the Z score;

[0086] Among them, the probability distribution model refers to a mathematical function or rule used to describe how a set of data or random variables are distributed. For example, normal distribution, binomial distribution, etc. The data parameter refers to the numerical characteristics used to fully determine the distribution in a specific probability distribution model. The probability density distribution function value refers to the density value calculated by the function for a specific random variable value in a probability density function. The Z score refers to the difference between the original score and the average.

[0087] Optionally, the probability distribution model of the abnormal data to be determined may be determined by hypothesis testing.

[0088] The embodiment of the present invention can more comprehensively evaluate the degree of abnormality of the data and improve the accuracy of abnormality detection by determining the multimodal abnormality probability of the abnormal data to be determined based on the single-mode abnormality probability. Wherein, the multimodal abnormality probability refers to the probability of abnormality or abnormal pattern occurrence in the data when processing multimodal data (i.e., data containing multiple types, such as text, images, sounds, etc.).

[0089] As an embodiment of the present invention, determining the multi-mode anomaly probability of the abnormal data to be determined according to the single-mode anomaly probability includes:

[0090] Analyzing the model performance of the probability distribution model corresponding to the single-mode abnormal probability;

[0091] determining the probability accuracy of the single-mode anomaly probability based on the model performance;

[0092] Determining a weight coefficient of the single-mode anomaly probability according to the probability accuracy;

[0093] According to the single-mode abnormal probability and the weight coefficient, the multi-mode abnormal probability of the abnormal data to be judged is calculated using the following formula:

[0094]

[0095] Among them, G drepresents the probability of multi-mode anomaly, n represents the number of single-mode anomaly probabilities, Q i represents the weight coefficient of the probability of the i-th single-mode anomaly, G i Represents the probability of single-mode anomaly.

[0096] The model performance refers to a series of indicators that determine the performance of a probability distribution model on a specific task. The probability accuracy refers to the accuracy of the model in predicting probabilities. The weight coefficient refers to the relative importance or influence value assigned to each single-mode anomaly probability.

[0097] The embodiment of the present invention can reduce the reliance on a single model and improve the robustness of the overall system to noise data and abnormal situations by identifying abnormal data in the abnormal data to be determined based on the multi-mode abnormal probability. The abnormal data refers to a group of data points in the data set that do not conform to the overall data pattern or distribution.

[0098] The bioelectric signal extraction module 104 is used to perform abnormal repair on the abnormal data to obtain repair data, define a weak signal extraction algorithm for the repair data, extract the bioelectric signal of the repair data based on the weak signal extraction algorithm, decompose the bioelectric signal to obtain a low-frequency component and a high-frequency component, and reconstruct the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal.

[0099] The embodiments of the present invention can significantly improve the overall quality of the data set by performing anomaly repair on the anomaly data to obtain repaired data, making the data more reliable and consistent. The repaired data refers to data obtained by processing and correcting the anomaly data in the original data set.

[0100] Optionally, as an embodiment of the present invention, the abnormal data is repaired abnormally to obtain the repaired data which can be repaired by database technology.

[0101] The embodiment of the present invention can effectively distinguish between signals and noise by defining a weak signal extraction algorithm for the repaired data, and can extract useful signals even when the signal is very weak and the noise interference is strong, thereby improving the quality and availability of the signal. The weak signal extraction algorithm refers to a type of algorithm specially designed to extract useful, weak signals in a strong noise background.

[0102] Optionally, as an embodiment of the present invention, the weak signal extraction algorithm for defining the repair data may be defined by a wavelet transform method.

[0103] The embodiment of the present invention can extract the bioelectric signal of the repair data based on the weak signal extraction algorithm, effectively suppress noise, improve the signal-to-noise ratio of the bioelectric signal, make the signal clearer, and facilitate subsequent analysis and processing. The bioelectric signal refers to the electrical signal generated by physiological activities in the body.

[0104] As an embodiment of the present invention, the step of extracting the bioelectric signal of the repair data based on the weak signal extraction algorithm includes:

[0105] Determining an average filter of the repaired data according to the repaired data;

[0106] Determining a matrix length and a filter angular frequency of the averaging filter;

[0107] The frequency characteristic of the average filter is calculated according to the matrix length and the filter angular frequency using the following formula:

[0108]

[0109] Among them, T x represents the frequency characteristic, L represents the matrix length, β represents the filter angular frequency, r represents the data signal position, sin represents the sine function, and e represents the exponential function with e as the base;

[0110] According to the frequency characteristics, the repair data is filtered using the average filter to obtain filtered data;

[0111] Based on the weak signal extraction algorithm, the bioelectric signal of the filtered data is extracted.

[0112] The average filter refers to a filter that smoothes a signal based on the average value of the calculated data points. The matrix length refers to the window size of the filter, that is, the number of sample points considered when performing the average calculation. The filter angular frequency refers to the frequency variable of the filter in the frequency domain analysis. The frequency characteristic refers to the response capability of the filter to signals of different frequencies. The filtered data refers to the data processed by the filter.

[0113] Optionally, the determination of the matrix length and filter angular frequency of the average filter may be determined by signal spectrum analysis.

[0114] The embodiment of the present invention can effectively improve the performance of signal processing and enhance the accuracy and reliability of the signal by decomposing the bioelectric signal to obtain low-frequency components and high-frequency components. The low-frequency component refers to the component with a lower frequency in the bioelectric signal. The high-frequency component refers to the component with a higher frequency in the bioelectric signal.

[0115] Optionally, as an embodiment of the present invention, the decomposition of the bioelectric signal to obtain the low-frequency component and the high-frequency component may be performed by Fourier transform method.

[0116] The embodiment of the present invention reconstructs the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal, which can specifically enhance the useful information in the signal while reducing noise and other interference, thereby improving the overall quality of the signal. The reconstructed bioelectric signal refers to a signal formed by processing the low-frequency component and the high-frequency component of the original bioelectric signal separately and then merging these processed components together.

[0117] As an embodiment of the present invention, reconstructing the bioelectric signal according to the low-frequency component and the high-frequency component to obtain the reconstructed bioelectric signal includes:

[0118] Analyzing the scaling function of the low-frequency component to determine a low-pass filter decomposition parameter of the scaling function;

[0119] Analyze the wavelet function of the high-frequency component to determine the high-pass filter decomposition parameters of the wavelet function;

[0120] Determining the number of decomposition layers, low-pass filter reconstruction parameters, and high-pass filter reconstruction parameters of the bioelectric signal;

[0121] According to the low-pass filter decomposition parameter, the high-pass filter decomposition parameter, the number of decomposition layers, the low-pass filter reconstruction parameter and the high-pass filter reconstruction parameter, the bioelectric signal is reconstructed using the following formula to obtain a reconstructed bioelectric signal:

[0122]

[0123] Among them, C j,q represents the reconstructed bioelectric signal, j represents the number of decomposition layers, m represents the summation variable, q represents the position index, S j Represents bioelectric signals, represents the low-pass filter reconstruction parameter, d represents the low-pass filter decomposition parameter, represents the high-pass filter reconstruction parameter, and g represents the high-pass filter decomposition parameter.

[0124] Among them, the scaling function refers to the mother function in the wavelet transform, which is used to generate low-frequency components. The wavelet function refers to another basic function in the wavelet transform, which is used to generate high-frequency components. The low-pass filter decomposition parameter refers to the coefficient of the filter used to extract the low-frequency component from the original signal. The high-pass filter decomposition parameter refers to the coefficient of the filter used to extract the high-frequency component from the original signal. The number of decomposition layers refers to the number of consecutive decompositions performed when the signal is subjected to multi-scale analysis in the wavelet transform. The low-pass filter reconstruction parameter refers to the filter coefficient used to reconstruct the low-frequency part of the original signal from the low-frequency approximation coefficients during the reconstruction process of the wavelet transform. The high-pass filter reconstruction parameter refers to the filter coefficient used to reconstruct the high-frequency part of the original signal from the high-frequency detail coefficients during the reconstruction process of the wavelet transform.

[0125] Optionally, the low-pass filter decomposition parameters for determining the scale function may be determined by a multi-scale analysis method.

[0126] The health index analysis module 105 is used to extract human health parameters from the reconstructed bioelectric signal, and analyze the health index of the target user based on the human health parameters.

[0127] The embodiment of the present invention can track the health status of an individual in real time by extracting the human health parameters of the reconstructed bioelectric signal, thereby improving the accuracy of health analysis. The human health parameters refer to quantitative indicators obtained by analyzing and reconstructing the bioelectric signal, which can reflect the physiological state and health status of the human body, such as heart rate, electrocardiogram waveform characteristics, blood pressure related parameters, etc.

[0128] The present invention can provide personalized health assessment by analyzing the health indicators of the target user based on the human health parameters, so as to better meet the differentiated health needs of individuals.

[0129] The embodiment of the present invention can comprehensively monitor multiple health indicators of the human body by configuring the sensor network of the target user, and provide a comprehensive health status assessment; optionally, the embodiment of the present invention can reduce the error or uncertainty that may exist in a single sensor by fusing the multi-sensor data based on the multi-analysis results, thereby improving the accuracy of the overall data; the embodiment of the present invention can ensure the stability and security of the system while improving the detection performance and more accurately identifying the abnormal mode by defining the abnormal detection algorithm of the fused sensor data; the embodiment of the present invention can more comprehensively evaluate the abnormal degree of the data by determining the multi-mode abnormal probability of the abnormal data to be judged according to the single-mode abnormal probability, thereby improving the accuracy of abnormal detection; the embodiment of the present invention can extract the bioelectric signal of the repair data based on the weak signal extraction algorithm, which can effectively suppress noise and improve the signal-to-noise ratio of the bioelectric signal, making the signal clearer and facilitating subsequent analysis and processing. Finally, the embodiment of the present invention can track the health status of the individual in real time by extracting the human health parameters of the reconstructed bioelectric signal, thereby improving the accuracy of health analysis. Therefore, the present invention can improve the comprehensiveness and accuracy of human health indicator status monitoring.

[0130] like Figure 2 FIG. 1 is a flow chart of a method for monitoring and analyzing the health status of human body indicators provided by an embodiment of the present invention. In this embodiment, the method for monitoring and analyzing the health status of human body indicators includes:

[0131] configuring a sensor network of a target user, collecting multi-sensor data of the target user in real time based on the sensor network, extracting feature vectors of the multi-sensor data, analyzing the multi-sensor data based on the feature vectors to obtain multi-analysis results, and fusing the multi-sensor data based on the multi-analysis results to obtain fused sensor data;

[0132] defining an anomaly detection algorithm for the fused sensor data, extracting window data of the fused sensor data using a trained sliding window model, detecting an outlier factor of the window data according to the anomaly detection algorithm, calculating an outlier coefficient of the window data based on the outlier factor, and classifying the window data corresponding to the outlier factor as abnormal data to be determined when the outlier coefficient is greater than a preset threshold;

[0133] Calculating the single-mode abnormality probability of the abnormal data to be determined, determining the multi-mode abnormality probability of the abnormal data to be determined according to the single-mode abnormality probability, and identifying the abnormal data in the abnormal data to be determined based on the multi-mode abnormality probability;

[0134] Performing abnormal repair on the abnormal data to obtain repaired data, defining a weak signal extraction algorithm for the repaired data, extracting a bioelectric signal of the repaired data based on the weak signal extraction algorithm, decomposing the bioelectric signal to obtain a low-frequency component and a high-frequency component, and reconstructing the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal;

[0135] Extract human health parameters of the reconstructed bioelectric signal, and analyze the health index of the target user based on the human health parameters.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A human health indicator status monitoring and analysis system, characterized in that: The system human health index status monitoring and analysis includes: a sensor data fusion module, an abnormality detection module, an abnormal data analysis module, a bioelectric signal extraction module and a health index analysis module; The fusion sensor data module is used to configure a sensor network of a target user, collect multi-sensor data of the target user in real time based on the sensor network, extract feature vectors of the multi-sensor data, analyze the multi-sensor data based on the feature vectors to obtain multi-analysis results, and fuse the multi-sensor data based on the multi-analysis results to obtain fused sensor data; The bioelectric signal extraction module is used to perform abnormal repair on the abnormal data to obtain repair data, define a weak signal extraction algorithm for the repair data, extract the bioelectric signal of the repair data based on the weak signal extraction algorithm, decompose the bioelectric signal to obtain a low-frequency component and a high-frequency component, and reconstruct the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal; The health index analysis module is used to extract human health parameters from the reconstructed bioelectric signal, and analyze the health index of the target user based on the human health parameters.

2. The human health indicator status monitoring and analysis system according to claim 1, characterized in that: The anomaly detection module is used to define an anomaly detection algorithm for the fused sensor data, use the trained sliding window model to extract the window data of the fused sensor data, detect the outlier factor of the window data according to the anomaly detection algorithm, calculate the outlier coefficient of the window data based on the outlier factor, and when the outlier coefficient is greater than a preset threshold, classify the window data corresponding to the outlier factor as abnormal data to be judged.

3. The human health indicator status monitoring and analysis system according to claim 2, characterized in that: The abnormal data analysis module is used to calculate the single-mode abnormality probability of the abnormal data to be judged, determine the multi-mode abnormality probability of the abnormal data to be judged according to the single-mode abnormality probability, and identify the abnormal data in the abnormal data to be judged based on the multi-mode abnormality probability.

4. The human health indicator status monitoring and analysis system according to claim 3, characterized in that: The extracting the feature vector of the multi-sensor data comprises: Determining feature extraction requirements for the multi-sensor data, and performing data cleaning on the multi-sensor data to obtain cleaned data; Analyzing the data types of the cleaned data, wherein the data types include: time series data, waveform data and image data; According to the data type, data features of the cleaned data are extracted.

5. The human health indicator status monitoring and analysis system according to claim 4, characterized in that: Calculating the feature correlation between the data feature and the feature extraction requirement, and analyzing the feature importance of the data feature using the trained feature importance analysis model; Based on the feature relevance and the feature importance, the data features are screened to obtain feature vectors.

6. The human health indicator status monitoring and analysis system according to claim 5, characterized in that: The step of extracting data features of the cleaned data according to the data type includes: Calculate the mean, variance and standard deviation of the cleaned data according to the time series data corresponding to the data type; The statistical characteristics of the cleaned data are determined according to the mean, the variance, and the standard deviation.

7. The human health indicator status monitoring and analysis system according to claim 6, characterized in that: Extracting waveform features of the cleaned data according to the waveform data corresponding to the data type; According to the image data corresponding to the data type, using a preset image feature extraction technology to extract image features of the cleaned data; The statistical features, the waveform features and the image features are integrated into data features.

8. The human health indicator status monitoring and analysis system according to claim 7, characterized in that: The step of fusing the multi-sensor data based on the multi-analysis results to obtain fused sensor data includes: Performing basic probability distribution on the multiple analysis results to obtain a probability distribution value; Normalizing the multi-sensor data to obtain multi-normalized sensor data; A data fusion algorithm for the multi-normalized sensor data is constructed according to the probability distribution value.

9. The human health indicator status monitoring and analysis system according to claim 8, characterized in that: Based on the data fusion algorithm, the multi-sensor data is fused to obtain initial fused data; The confidence of the initial fusion data is calculated, and when the confidence meets a preset confidence interval, the initial fusion data is used as the fusion sensing data of the multi-sensing data.

10. A method for monitoring and analyzing human health indicators, the method being implemented based on claim 3, characterized in that: The method comprises: configuring a sensor network of a target user, collecting multi-sensor data of the target user in real time based on the sensor network, extracting feature vectors of the multi-sensor data, analyzing the multi-sensor data based on the feature vectors to obtain multi-analysis results, and fusing the multi-sensor data based on the multi-analysis results to obtain fused sensor data; defining an anomaly detection algorithm for the fused sensor data, extracting window data of the fused sensor data using a trained sliding window model, detecting an outlier factor of the window data according to the anomaly detection algorithm, calculating an outlier coefficient of the window data based on the outlier factor, and classifying the window data corresponding to the outlier factor as abnormal data to be determined when the outlier coefficient is greater than a preset threshold; Calculating the single-mode abnormality probability of the abnormal data to be determined, determining the multi-mode abnormality probability of the abnormal data to be determined according to the single-mode abnormality probability, and identifying the abnormal data in the abnormal data to be determined based on the multi-mode abnormality probability; Performing abnormal repair on the abnormal data to obtain repaired data, defining a weak signal extraction algorithm for the repaired data, extracting a bioelectric signal of the repaired data based on the weak signal extraction algorithm, decomposing the bioelectric signal to obtain a low-frequency component and a high-frequency component, and reconstructing the bioelectric signal according to the low-frequency component and the high-frequency component to obtain a reconstructed bioelectric signal; Extract human health parameters of the reconstructed bioelectric signal, and analyze the health index of the target user based on the human health parameters.

Images