Method and system for monitoring and analyzing behavior data of middle-aged and elderly people in real time in intelligent health care

Abstract

The invention relates to the technical field of intelligent health, and discloses an intelligent health middle-aged and elderly people behavior data real-time monitoring and analysis method and system, and the method comprises the steps: collecting a biological resonance signal, a gas spectrum and a sound wave signal of middle-aged and elderly people; constructing a resonance characteristic spectrum of the middle-aged and elderly people, and analyzing physiological feedback characteristics of the middle-aged and elderly people by using the resonance characteristic spectrum; constructing an odor-behavior association map of the middle-aged and elderly people, and identifying gas feedback characteristics of the middle-aged and elderly people by using the odor-behavior association map; performing frequency domain decomposition on the sound wave signal to obtain an infrasonic frequency band and an ultrasonic frequency band, identifying internal organ vibration characteristics and body surface action characteristics of the middle-aged and elderly people, and analyzing voiceprint feedback characteristics of the middle-aged and elderly people; the physiological feedback features, the gas feedback features and the voiceprint feedback features are used for conducting behavior analysis on the middle-aged and elderly people, and a behavior monitoring report is obtained. The method can improve the reliability of behavior abnormality analysis of the elderly in smart health.

Description

Technical Field

[0001] The present invention relates to a method and system for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care, and belongs to the field of smart health care technology. Background Art

[0002] Smart healthcare refers to a new model that leverages advanced technologies to integrate medical, elderly care, and health management services to provide seniors with intelligent, personalized, efficient, and convenient healthcare services. Real-time monitoring and analysis of behavioral data from middle-aged and elderly individuals can identify abnormal behaviors and provide more personalized and efficient health management services. This approach plays a significant role in promoting the development of the smart healthcare industry and alleviating the burden of elderly care on society.

[0003] Currently, behavioral data monitoring for middle-aged and elderly individuals is often based on a combination of wearable devices and fixed sensors. For example, wearable devices like smart bracelets and smartwatches collect physiological data such as heart rate and step count, while infrared sensors and cameras are installed indoors to monitor the elderly's movements. This approach can quickly identify obvious abnormalities in middle-aged and elderly individuals, such as rapid increases in blood pressure and falls. However, traditional solutions lack the ability to fuse multi-source data, making it difficult for different types of data to verify each other. Furthermore, deviations in single sensor data or environmental interference can lead to false alarms, resulting in low reliability in analyzing behavioral anomalies in smart healthcare for the elderly. Summary of the Invention

[0004] The present invention provides a method and system for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care, the main purpose of which is to improve the reliability of abnormal behavior analysis of middle-aged and elderly people in smart health care.

[0005] To achieve the above objectives, the present invention provides a method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care, including:

[0006] Querying the behavior time information of the middle-aged and elderly people in the smart health care environment, and collecting the bioresonance signals, gas spectra and sound wave signals of the middle-aged and elderly people based on the behavior time information;

[0007] performing tensor field decomposition on the bioresonance signal to obtain a decomposed signal, constructing a resonance characteristic spectrum of the middle-aged and elderly people using the decomposed signal, and analyzing the physiological feedback characteristics of the middle-aged and elderly people using the resonance characteristic spectrum;

[0008] performing temperature compensation on the gas spectrum to obtain a corrected gas spectrum, performing feature compression on the corrected gas spectrum to obtain a gas feature vector, constructing an odor-behavior association map of the middle-aged and elderly people using the gas feature vector, and identifying the gas feedback characteristics of the middle-aged and elderly people using the odor-behavior association map;

[0009] Decomposing the acoustic wave signal in the frequency domain to obtain an infrasonic frequency band and an ultrasonic frequency band, using the infrasonic frequency band to identify the vibration characteristics of the internal organs of the middle-aged and elderly people, and using the ultrasonic frequency band to identify the surface motion characteristics of the middle-aged and elderly people, spatiotemporally fusing the internal organ vibration characteristics and the surface motion characteristics to obtain a multimodal voiceprint dataset, and using the multimodal voiceprint dataset to analyze the voiceprint feedback characteristics of the middle-aged and elderly people;

[0010] The physiological feedback characteristics, the gas feedback characteristics and the voiceprint feedback characteristics are used to perform behavior analysis on the middle-aged and elderly people to obtain a behavior monitoring report.

[0011] Optionally, based on the behavior time information, collecting the bioresonance signal, gas spectrum and sound wave signal of the middle-aged and elderly people includes:

[0012] Based on the behavior time information, dividing the behavior time axis of the middle-aged and elderly people to obtain behavior time segments;

[0013] Based on the behavior time segments, deploying trimodal sensors in the activity areas of the middle-aged and elderly people;

[0014] The trimodal sensor is used to collect the biological resonance signals, gas spectra and sound wave signals of the middle-aged and elderly people.

[0015] Optionally, performing tensor field decomposition on the bioresonance signal to obtain a decomposed signal includes:

[0016] performing quantum noise suppression on the bioresonance signal to obtain a noise-reduced three-dimensional tensor;

[0017] Performing high-order tensor decomposition on the denoised three-dimensional tensor to obtain a core tensor and a factor matrix group;

[0018] Performing physiological semantic mapping on the factor matrix group to obtain a physiological mapping tensor;

[0019] performing joint tensor slicing processing on the core tensor and the physiological mapping tensor to obtain a physiological component tensor;

[0020] The physiological component tensor is subjected to spatiotemporal reconstruction to obtain a decomposed signal.

[0021] Optionally, the resonance characteristic spectrum is used to analyze physiological feedback characteristics of the middle-aged and elderly people, including:

[0022] querying historical resonance characteristic data of the middle-aged and elderly people to construct a standard reference atlas of the middle-aged and elderly people;

[0023] Extracting key characteristic parameters of the same latitude from the resonance characteristic spectrum and the standard reference spectrum to obtain comparison parameters;

[0024] quantifying the difference of the comparison parameters to obtain a quantitative difference value;

[0025] Based on the quantified difference value, abnormal marking is performed on the resonance characteristic spectrum to obtain a marked characteristic point;

[0026] The physiological feedback characteristics of the middle-aged and elderly people are identified by using the marked feature points.

[0027] Optionally, using the gas feature vectors to construct an odor-behavior association map for the middle-aged and elderly, including:

[0028] Using the gas characteristic vector, constructing the behavior synchronization gas matrix of the middle-aged and elderly people;

[0029] performing a graph node embedding operation on the behavior synchronization gas matrix to obtain an odor behavior node;

[0030] Using the odor behavior nodes, constructing a behavioral dynamic graph of the middle-aged and elderly people;

[0031] Performing heterogeneous graph convolution processing on the behavior dynamic graph to obtain behavior correlation features;

[0032] The behavior association features are subjected to graph visualization processing to obtain an odor-behavior association graph.

[0033] Optionally, using the odor-behavior association map to identify the gas feedback characteristics of the middle-aged and elderly people includes:

[0034] Performing key point screening on the odor-behavior association map to obtain a target behavior focus group;

[0035] Based on the target behavior focus group, identifying the risk propagation link of the odor-behavior association graph;

[0036] Utilizing the risk propagation link, the target focus area of ​​the middle-aged and elderly people is located to obtain feedback areas;

[0037] Performing risk rating on the feedback parts to obtain a health risk level table;

[0038] The health risk level table is used to identify the gas feedback characteristics of the middle-aged and elderly people.

[0039] Optionally, based on the target behavior focus group, identifying the risk propagation link of the odor-behavior association graph includes:

[0040] Calculate the risk propagation probability of key points in the target behavior focus group:

[0041]

[0042] Among them, P i→j represents the risk propagation probability, i represents the key point i in the target behavior focus group, j represents the key point j in the target behavior focus group, i→j represents a link from node i to node j, W i，j represents the edge weight from key point i to j, N(i) represents the set of leading points of key point i, W i，k Represents the weight of key point i to adjacent point k;

[0043] Based on the risk propagation probability, target links are filtered from the target behavior focus group to obtain risk propagation links.

[0044] Optionally, the internal organ vibration features and the body surface motion features are temporally and spatially fused to obtain a multimodal voiceprint dataset, including:

[0045] Performing a time alignment operation on the internal organ vibration feature and the body surface motion feature to obtain a synchronized spatiotemporal feature;

[0046] Performing action mapping on the synchronous spatiotemporal features to obtain a synchronous feature matrix;

[0047] Performing feature fusion on the synchronized feature matrix to obtain a multimodal action vector;

[0048] Performing spatiotemporal encoding on the multimodal motion vector to obtain a spatiotemporal motion feature set;

[0049] The dataset of the spatiotemporal action feature set is reconstructed to obtain a multimodal voiceprint dataset.

[0050] Optionally, the physiological feedback characteristics, the gas feedback characteristics, and the voiceprint feedback characteristics are used to perform behavioral analysis on the middle-aged and elderly people to obtain a behavioral monitoring report, including:

[0051] Using the physiological feedback characteristics, assessing the physiological risk level of the middle-aged and elderly people;

[0052] Using the gas feedback characteristics, scoring the respiratory function of the middle-aged and elderly people to obtain a respiratory score;

[0053] Calculating the voiceprint health index of the middle-aged and elderly person using the voiceprint feedback feature;

[0054] Based on the physiological risk level, the breathing score and the voiceprint health index, the middle-aged and elderly people are warned of abnormal behavior events to obtain warning information;

[0055] Based on the early warning information, a behavioral recommendation plan for the middle-aged and elderly people is constructed to obtain a behavior monitoring report.

[0056] In order to solve the above problems, the present invention also provides a real-time monitoring and analysis system for behavioral data of middle-aged and elderly people in smart health care, which includes:

[0057] A signal acquisition module is used to query the behavior time information of the middle-aged and elderly people in the smart health care environment, and based on the behavior time information, collect the biological resonance signal, gas spectrum and sound wave signal of the middle-aged and elderly people;

[0058] a physiological characteristic analysis module, configured to perform tensor field decomposition on the bioresonance signal to obtain a decomposed signal, construct a resonance characteristic spectrum of the middle-aged and elderly people using the decomposed signal, and analyze the physiological feedback characteristics of the middle-aged and elderly people using the resonance characteristic spectrum;

[0059] a gas characteristic analysis module, configured to perform temperature compensation on the gas spectrum to obtain a corrected gas spectrum, perform feature compression on the corrected gas spectrum to obtain a gas characteristic vector, construct an odor-behavior association map of the middle-aged and elderly using the gas characteristic vector, and identify the gas feedback characteristics of the middle-aged and elderly using the odor-behavior association map;

[0060] an acoustic feature analysis module for performing frequency domain decomposition of the acoustic wave signal to obtain infrasonic frequency bands and ultrasonic frequency bands, identifying the vibration characteristics of the internal organs of the middle-aged and elderly individuals using the infrasonic frequency bands, identifying the surface motion characteristics of the middle-aged and elderly individuals using the ultrasonic frequency bands, performing spatiotemporal fusion of the internal organ vibration characteristics and the surface motion characteristics to obtain a multimodal voiceprint dataset, and analyzing the voiceprint feedback characteristics of the middle-aged and elderly individuals using the multimodal voiceprint dataset;

[0061] The comprehensive feature analysis module is used to use the physiological feedback features, the gas feedback features and the voiceprint feedback features to perform behavioral analysis on the middle-aged and elderly people and obtain a behavior monitoring report.

[0062] Compared with the problems described in the background technology, the behavioral time information of middle-aged and elderly people (such as exercise, lunch break, and sleep period) is obtained, and divided into dynamic activities, static activities, sleep and other behavioral time segments according to the time axis (such as dynamic activities from 7 to 9 am and static activities from 3 to 5 pm). Trimodal sensors are deployed according to different time periods: bioresonance sensors are used to collect organ vibration signals during dynamic activities, gas spectrometers are used to collect respiratory metabolic gases during static activities, and acoustic sensors are used to capture body surface movements and sound signals at all times. This can avoid redundant sensor work in unnecessary time periods (such as turning off motion sensors during sleep), reduce the amount of invalid data, and increase the proportion of valid data. Furthermore, the present invention can identify the physiological characteristics of middle-aged and elderly people by decomposing bioresonance signals, which can detect physiological changes early. Abnormality; further, the present invention constructs an odor-behavior correlation map by utilizing gas spectra to identify the gas feedback characteristics of middle-aged and elderly people, thereby intuitively discovering the causal chain of "gas characteristics → physiological abnormalities → behavioral abnormalities" (such as increased acetone → liver metabolism → unstable gait), and improving the accuracy of risk positioning; further, the present invention decomposes the sound wave signal into infrasound and ultrasound to identify the internal organ vibration characteristics and surface movement characteristics of middle-aged and elderly people, thereby identifying the voiceprint feedback characteristics of middle-aged and elderly people; further, the present invention realizes accurate monitoring and early warning of the physical condition of middle-aged and elderly people through multi-dimensional feature fusion, such as by analyzing abnormal heart rate, changes in respiratory gas composition and voice fundamental frequency fluctuations, timely discovering abnormal behavioral tendencies such as dizziness and fatigue caused by hypoglycemia in the elderly. Therefore, the present invention can improve the reliability of behavioral abnormality analysis of middle-aged and elderly people in smart health care. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A flowchart of a method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care provided by one embodiment of the present invention;

[0064] Figure 2 A schematic diagram of a module for implementing the method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care provided by one embodiment of the present invention.

[0065] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0066] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0067] The embodiment of the present application provides a method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care. The execution subject of the method includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care can be executed by software or hardware installed on a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0068] Example 1:

[0069] Reference Figure 1 The figure is a flow chart of a method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to one embodiment of the present invention. In this embodiment, the method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care includes:

[0070] S1. Query the behavior time information of the middle-aged and elderly people in the smart health care environment, and based on the behavior time information, collect the biological resonance signals, gas spectra and sound wave signals of the middle-aged and elderly people.

[0071] The embodiment of the present invention can understand the specific time points when the middle-aged and elderly people perform activities or rest by querying the behavioral time information of the middle-aged and elderly people in the smart health care environment, so as to better deploy the collection time of the sensors. For example, motion-related sensors will not be deployed when the middle-aged and elderly people are resting, so as to improve the efficiency of data collection.

[0072] Among them, the behavior time information refers to the specific time or time period of the behavior of middle-aged and elderly people, such as 6 to 7 in the morning is the exercise time for middle-aged and elderly people, and 1 to 2 in the afternoon is the lunch break time for middle-aged and elderly people.

[0073] Optionally, the behavior time information can be obtained by querying the behavior activity time data of middle-aged and elderly people recorded in the data center of the smart health care service platform.

[0074] Furthermore, the embodiments of the present invention collect the biological resonance signals, gas spectra and sound wave signals of the middle-aged and elderly people based on the behavioral time information, and can deploy different sensors in different behavioral time periods of the middle-aged and elderly people to collect corresponding behavioral data, thereby improving the targeted analysis of the behavior of the middle-aged and elderly people, such as collecting vibration signals during exercise and collecting gas signals during rest.

[0075] Among them, the bioresonance signal refers to the weak electromagnetic vibration signal generated by human tissues and organs during physiological activities, which is collected by the bioresonance sensor; the gas spectrum refers to the gas discharged during human breathing and metabolism, which is collected by the gas spectrometer; and the sound wave signal refers to the sound wave signal generated and vibrated by the human body, which is collected by the sound wave sensor.

[0076] As an embodiment of the present invention, based on the behavior time information, collecting the bioresonance signal, gas spectrum and sound wave signal of the middle-aged and elderly people includes:

[0077] Based on the behavior time information, dividing the behavior time axis of the middle-aged and elderly people to obtain behavior time segments;

[0078] Based on the behavior time segments, deploying trimodal sensors in the activity areas of the middle-aged and elderly people;

[0079] The trimodal sensor is used to collect the biological resonance signals, gas spectra and sound wave signals of the middle-aged and elderly people.

[0080] Among them, the behavioral time segments refer to dividing continuous behaviors into short time periods with specific activity attributes based on the daily activity patterns of middle-aged and elderly people (such as eating, walking, resting, etc.) through time axis annotation technology. The trimodal sensor refers to a bioresonance sensor, a gas spectrometer and an acoustic wave sensor.

[0081] During the specific implementation process, the activity time of middle-aged and elderly people can be divided into dynamic activity time, static activity and sleep behavior segments. For example, the dynamic activities (walking, stretching) performed by middle-aged and elderly people from 7 to 9 in the morning can be divided into dynamic activity time, the static activities (playing chess, playing cards) performed by middle-aged and elderly people from 3 to 5 in the afternoon, and the sleep time of middle-aged and elderly people from 22:00 to 6:00 in the evening need to be divided in combination with the data of the time survey; during the dynamic activity time, bioresonance sensors are deployed to collect the weak electromagnetic vibration signals generated by the human tissues and organs of middle-aged and elderly people during physiological activities; during the static activity time, acoustic wave sensors are deployed to collect the acoustic wave signals generated and vibrated by the human body of middle-aged and elderly people; during the static activity time, a gas spectrometer is used to collect the gas spectrum of middle-aged and elderly people.

[0082] S2. Performing tensor field decomposition on the bioresonance signal to obtain a decomposed signal, constructing a resonance characteristic spectrum of the middle-aged and elderly people using the decomposed signal, and analyzing the physiological feedback characteristics of the middle-aged and elderly people using the resonance characteristic spectrum.

[0083] The embodiments of the present invention perform tensor field decomposition on the bioresonance signal to obtain a decomposed signal, which can split the mixed original signal into simple components and quickly discover hidden feature changes in the signal. This is like breaking a complex puzzle into small pieces, making it easier to see whether the puzzle pattern has changed, thereby facilitating timely detection of abnormalities in the human physiological state.

[0084] The decomposed signal refers to decomposing the complex original bioresonance signal into multiple low-dimensional simple signal components with specific frequency, time and space characteristics through tensor field decomposition technology.

[0085] As an embodiment of the present invention, performing tensor field decomposition on the bioresonance signal to obtain a decomposed signal includes:

[0086] performing quantum noise suppression on the bioresonance signal to obtain a noise-reduced three-dimensional tensor;

[0087] Performing high-order tensor decomposition on the denoised three-dimensional tensor to obtain a core tensor and a factor matrix group;

[0088] Performing physiological semantic mapping on the factor matrix group to obtain a physiological mapping tensor;

[0089] performing joint tensor slicing processing on the core tensor and the physiological mapping tensor to obtain a physiological component tensor;

[0090] The physiological component tensor is subjected to spatiotemporal reconstruction to obtain a decomposed signal.

[0091] Among them, the core tensor refers to the low-dimensional tensor obtained after the decomposition of the high-order tensor, the factor matrix group refers to a group of pattern matrices generated by the decomposition of the high-order tensor, including the time factor matrix, the space factor matrix and the frequency factor matrix, which correspond to the three dimensions of the original tensor respectively, and the physiological mapping tensor refers to the conversion of the factor matrix group into a tensor with physiological significance through physiological semantic mapping (such as predefined organ vibration frequency range and spatial position), for example, mapping the 0.8-1.2Hz component in the frequency factor matrix to the "heart vibration feature tensor", and mapping the component corresponding to the chest area sensor in the spatial factor matrix to the "lung motion feature tensor".

[0092] During the specific implementation process, the bioresonance signal can be quantum noise suppressed, and the background noise can be removed while maintaining the quantum coherence of the signal through quantum enhanced wavelet transform combined with an adaptive threshold algorithm (such as Bayes Shrink), thereby obtaining a denoised three-dimensional tensor with dimensions of time × space × frequency; the denoised three-dimensional tensor is subjected to high-order tensor decomposition, and Tucker decomposition can be used to decompose the tensor into a core tensor and three factor matrix groups (time pattern matrix, space pattern matrix and frequency pattern matrix); the factor matrix group can be physiologically semantically mapped through a trained physiological feature dictionary, such as mapping the frequency factor matrix to the physiological feature tensor of organs such as the heart and lungs; a sliding window slice is performed every 5 seconds along the time axis to extract spatiotemporal feature blocks related to organ function to obtain a physiological component tensor; a spatiotemporal convolutional neural network (STCNN) is used in combination with a variational autoencoder (VAE) to reconstruct independent dynamic bioresonance signals of each organ (such as heart vibration waveform, respiratory movement trajectory) based on the slice features to obtain a decomposed signal.

[0093] Furthermore, the embodiment of the present invention utilizes the decomposed signal to construct the resonance characteristic map of the middle-aged and elderly people. The key features of the decomposed signal can be integrated into a visual chart according to specific rules to intuitively present the laws and characteristics of their body resonance.

[0094] The resonance characteristic map refers to a collection of individual or group resonance patterns and physiological state characteristics that are presented in a visual chart by integrating the decomposition characteristics of the biological resonance signal (such as frequency, intensity or spatiotemporal distribution, etc.).

[0095] Optionally, the resonance feature map is obtained by extracting key features of the decomposed signal and associating them with organs, and constructing a visual map of nodes (organs) and edges (feature associations) in a graph structure.

[0096] Furthermore, the embodiments of the present invention can intuitively and accurately identify changes in the functional status of the internal organs and tissues of the middle-aged and elderly people by utilizing the resonance characteristic map to analyze the physiological feedback characteristics of the middle-aged and elderly people, and discover potential health risks in advance. For example, by comparing the resonance signal characteristics of the cardiopulmonary region in the map, signs of cardiopulmonary function decline can be detected in time, providing a key basis for early detection and early intervention of diseases.

[0097] The physiological feedback characteristics refer to the detectable and analyzable specific indicators or laws related to the functional state of human organs, tissues or systems expressed through biological signals during physiological activities.

[0098] As an embodiment of the present invention, analyzing the physiological feedback characteristics of the middle-aged and elderly people by using the resonance characteristic spectrum includes:

[0099] querying historical resonance characteristic data of the middle-aged and elderly people to construct a standard reference atlas of the middle-aged and elderly people;

[0100] Extracting key characteristic parameters of the same latitude from the resonance characteristic spectrum and the standard reference spectrum to obtain comparison parameters;

[0101] quantifying the difference of the comparison parameters to obtain a quantitative difference value;

[0102] Based on the quantified difference value, abnormal marking is performed on the resonance characteristic spectrum to obtain a marked characteristic point;

[0103] The physiological feedback characteristics of the middle-aged and elderly people are identified by using the marked feature points.

[0104] Among them, the standard reference atlas refers to an individualized normal physiological characteristic atlas constructed based on the historical resonance characteristic data of middle-aged and elderly people (such as the statistical distribution of the resonance frequency and amplitude of each organ in the past period of time), and the quantitative difference value refers to the amplitude of the characteristic parameter deviating from the normal range, which is used to quantitatively judge whether the physiological state is abnormal.

[0105] During the specific implementation process, historical resonance feature data of middle-aged and elderly people can be retrieved (such as the mean organ resonance frequency and amplitude in the past three months), and statistical modeling (such as the mean-standard deviation ellipse) can be used to construct their individual standard reference map; key feature parameters of the same dimension (such as the spatiotemporal distribution of heart vibration frequency and lung resonance amplitude) can be synchronously extracted from the current resonance feature map and the standard reference map, and the parameter dimensions (time, space and frequency) can be ensured to be completely aligned; the difference between the current feature parameters and the standard reference values ​​can be calculated using measurement methods such as Euclidean distance and cosine similarity, and the quantitative difference value can be obtained through Z-score normalization (such as heart frequency difference value = (current value - historical mean) / historical standard deviation); a difference threshold can be set (such as Z-score > 2), and feature parameters that exceed the threshold can be visually marked in the map (such as red highlighted nodes or bold edges) to generate a set of marked feature points; based on the distribution of marked feature points (such as abnormal increase in heart vibration frequency in three consecutive time segments) and combined with the physiological knowledge map (such as the association rules between frequency increase and myocardial ischemia), the physiological feedback characteristics of middle-aged and elderly people (such as abnormal cardiac load) can be identified.

[0106] S3. Perform temperature compensation on the gas spectrum to obtain a corrected gas spectrum, perform feature compression on the corrected gas spectrum to obtain a gas feature vector, use the gas feature vector to construct an odor-behavior association map of the middle-aged and elderly people, and use the odor-behavior association map to identify the gas feedback characteristics of the middle-aged and elderly people.

[0107] The embodiment of the present invention performs temperature compensation on the gas spectrum to obtain a corrected gas spectrum, which can eliminate the interference of ambient temperature changes on the detection results of odor molecules, and ensure that the spectral data truly reflects the characteristics of volatile organic compounds emitted by the human body. For example, it can avoid the false increase in the spectral intensity of a certain type of VOCs due to increased room temperature, thereby avoiding misjudging the health status of middle-aged and elderly people.

[0108] In the specific implementation process, a multivariate regression model of temperature and gas spectral absorption coefficient (such as y = a·T 2 +b·T+c, where y is the absorption coefficient and T is the temperature), the spectral baseline is dynamically adjusted based on the real-time temperature sensor data to eliminate the spectral line drift caused by temperature fluctuations (such as ±5°C) and obtain the corrected gas spectrum.

[0109] The embodiment of the present invention performs feature compression on the calibration gas spectrum to obtain a gas feature vector, which can simplify high-dimensional and complex spectral data into low-dimensional key features, thereby improving data processing efficiency and retaining core information.

[0110] In the specific implementation process, principal component analysis (PCA) can be used to perform feature compression on the calibration gas spectrum, extract the top N principal components (e.g., N=10) with a cumulative variance contribution rate of more than 95%, and map the high-dimensional spectral data (e.g., 1000+ wavelength points) into low-dimensional gas feature vectors.

[0111] Furthermore, the embodiment of the present invention utilizes the gas characteristic vectors to construct the odor-behavior association map of the middle-aged and elderly people, which can intuitively present the potential connection between human body odor and behavior, and explore the association patterns between odor changes and behavioral activities.

[0112] Among them, the odor-behavior association map refers to a mapping relationship map between odor characteristics and behavioral patterns constructed by collecting and analyzing the chemical characteristics of specific odor molecules (such as volatile organic compound composition and concentration distribution) and combining individual behavioral data (such as emotional reactions, changes in physiological indicators and movement trajectories, etc.).

[0113] As an embodiment of the present invention, the gas feature vector is used to construct an odor-behavior association map of the middle-aged and elderly people, including:

[0114] Using the gas characteristic vector, constructing the behavior synchronization gas matrix of the middle-aged and elderly people;

[0115] performing a graph node embedding operation on the behavior synchronization gas matrix to obtain an odor behavior node;

[0116] Using the odor behavior nodes, constructing a behavioral dynamic graph of the middle-aged and elderly people;

[0117] Performing heterogeneous graph convolution processing on the behavior dynamic graph to obtain behavior correlation features;

[0118] The behavior association features are subjected to graph visualization processing to obtain an odor-behavior association graph.

[0119] The behavior synchronization gas matrix is:

[0120] During the specific implementation process, the gas feature vectors under the same behavioral time segment can be stacked in rows, and the columns represent the characteristics of different gas components to construct a behavioral synchronized gas matrix with the dimension of "behavior-gas"; the node embedding algorithm in the graph neural network (such as DeepWalk) is used to map each row in the behavioral synchronized gas matrix (corresponding to a behavior and related gas characteristics) into a low-dimensional vector as the odor behavior node in the graph; with the odor behavior node as the vertex, based on the behavioral time series or gas feature similarity, directed edges are added by setting a threshold (such as cosine similarity > 0.7) to form a behavioral dynamic graph that reflects the dynamic relationship between the behavior and gas characteristics of middle-aged and elderly people; a heterogeneous graph convolutional network (such as HGCN) is used to extract features from the behavioral dynamic graph, aggregate the neighborhood information of different types of nodes (behavior nodes, gas nodes), learn the high-order correlation between nodes, and obtain behavioral correlation features; using a graph visualization tool (such as Gephi), the behavioral correlation features are mapped to node size, color and edge thickness, and annotation labels are added to intuitively present the correlation between the gas characteristics and behavioral patterns of middle-aged and elderly people, forming an odor-behavior correlation graph.

[0121] Furthermore, the embodiment of the present invention utilizes the odor-behavior association map to identify the gas feedback characteristics of the middle-aged and elderly people. Through the correspondence between odor and behavior in the map, the odor signal characteristics related to specific behavior or health status can be accurately extracted to provide a basis for analyzing the physical state. For example, abnormal odor characteristics of the elderly during sleep can be quickly identified from the map to assist in judging sleep quality or potential health problems.

[0122] As an embodiment of the present invention, the gas feedback characteristics of the middle-aged and elderly people are identified by using the odor-behavior association map, including:

[0123] Performing key point screening on the odor-behavior association map to obtain a target behavior focus group;

[0124] Based on the target behavior focus group, identifying the risk propagation link of the odor-behavior association graph;

[0125] Utilizing the risk propagation link, the target focus area of ​​the middle-aged and elderly people is located to obtain feedback areas;

[0126] Performing risk rating on the feedback parts to obtain a health risk level table;

[0127] The health risk level table is used to identify the gas feedback characteristics of the middle-aged and elderly people.

[0128] Among them, the risk propagation link refers to a directed path in the odor-behavior association map, from the odor feature node (such as abnormal volatile organic compound concentration) through several intermediate nodes (such as metabolic abnormalities, fluctuations in physiological indicators) to the behavior abnormality node (such as falls, difficulty breathing), reflecting the potential causal relationship chain from changes in gas composition to behavioral abnormalities (such as "exhaled NO concentration ↑→airway inflammation node→coughing behavior node→decreased balance ability node"). The health risk level table refers to a table generated through quantitative evaluation based on the structural characteristics of the risk propagation link (such as path length, edge weight) and the degree of node abnormality.

[0129] During the specific implementation process, the node importance score can be calculated through the graph attention mechanism (GAT), and the top 20% of nodes with the preset target behavior (such as falls and dyspnea) can be screened out to form the target behavior attention group; according to the physiological sources corresponding to the gas components in the transmission link (such as acetone comes from liver metabolism), the feedback site (such as liver, pancreas) is located in combination with anatomical knowledge, and the association strength is quantified by the graph edge weight (such as edge weight 0.8 indicates strong association); based on the link length, node abnormality level (such as Z-score value) and medical knowledge base, the hierarchical analysis method is used to assign risk weights to the feedback site and generate a health risk level table (such as liver metabolism risk: high, medium, low); the health risk level table is compared with the gas feature vector, and features exceeding the threshold are extracted (such as acetone concentration Z-score>2.5). Combined with the site positioning results, the gas feedback features are identified (such as "abnormal liver metabolism leads to increased exhaled acetone").

[0130] Preferably, the step of identifying the risk propagation link of the odor-behavior association graph based on the target behavior focus group includes:

[0131] Calculate the risk propagation probability of key points in the target behavior focus group:

[0132]

[0133] Among them, P i→j represents the risk propagation probability, i represents the key point i in the target behavior focus group, j represents the key point j in the target behavior focus group, i→j represents a link from node i to node j, W i，j represents the edge weight from key point i to j, N(i) represents the set of leading points of key point i, W i，k Represents the weight of key point i to adjacent point k;

[0134] Based on the risk propagation probability, target links are filtered from the target behavior focus group to obtain risk propagation links.

[0135] During the specific implementation process, based on the risk propagation probability matrix, a dynamic programming algorithm can be used to start from the starting node (such as the abnormal gas characteristic node) and select links with a cumulative probability greater than 0.7 and a path length ≤ 3 as risk propagation links.

[0136] It should be further explained that the risk propagation probability calculation formula is used to quantify the possibility of risk propagation between key points in the odor-behavior association map, and calculate the probability value of risk transmission between nodes through edge weights, providing data support for screening high-credible risk propagation links. For example, it can identify the probability path of "increased exhaled breath characteristic → metabolic abnormality → behavioral abnormality", and assist in judging the strength of the causal relationship between gas characteristics and behavioral abnormalities.

[0137] S4. Decompose the sound wave signal in the frequency domain to obtain an infrasonic frequency band and an ultrasonic frequency band, use the infrasonic frequency band to identify the vibration characteristics of the internal organs of the middle-aged and elderly people, use the ultrasonic frequency band to identify the surface movement characteristics of the middle-aged and elderly people, and perform spatiotemporal fusion of the internal organ vibration characteristics and the surface movement characteristics to obtain a multimodal voiceprint dataset. Use the multimodal voiceprint dataset to analyze the voiceprint feedback characteristics of the middle-aged and elderly people.

[0138] The embodiment of the present invention decomposes the sound wave signal in the frequency domain to obtain the infrasonic frequency band and the ultrasonic frequency band, which can respectively capture low-frequency physiological activity signals (such as organ vibration) and high-frequency environmental detail signals, thereby comprehensively analyzing the hidden information in the sound wave.

[0139] The infrasonic frequency band refers to a sound wave frequency band with a frequency range lower than 20 Hz, and the ultrasonic frequency band refers to a sound wave frequency band with a frequency range higher than 20 kHz.

[0140] During the specific implementation process, the fast Fourier transform algorithm can be used to perform spectrum analysis on the sound wave signal, and the signal can be decomposed into the corresponding frequency band by setting the frequency threshold (infrasonic frequency band <20Hz, ultrasonic frequency band >20kHz).

[0141] Furthermore, the embodiments of the present invention can identify the vibration characteristics of the internal organs of the middle-aged and elderly people by utilizing the infrasound frequency band, and can capture the abnormal vibration patterns of the organs through low-frequency signals, and detect signs of organ function decline or pathology in advance. For example, by analyzing the abnormal fluctuations in the liver vibration frequency in infrasound, it can assist in determining whether the liver has inflammation or fibrosis tendencies.

[0142] Among them, the internal organ vibration characteristics refer to the specific parameters such as frequency, amplitude, phase and waveform pattern of the infrasound frequency band vibrations generated by the internal organs of the middle-aged and elderly people (such as the heart, lungs and gastrointestinal tract) under normal physiological activities or pathological conditions.

[0143] During the specific implementation process, the vibration characteristics of internal organs such as the heart and lungs of middle-aged and elderly people can be identified by matching the infrasound frequency band (such as 0.8-2.5Hz) with the preset organ vibration frequency template (such as 0.8-1.2Hz for the heart and 1.5-2.5Hz for the lungs), using wavelet packet decomposition to extract time-frequency features, and combining independent component analysis (ICA) to separate aliasing signals.

[0144] Furthermore, the embodiment of the present invention utilizes the ultrasonic frequency band to identify the surface movement characteristics of the middle-aged and elderly people, and can accurately capture the air vibration or surface displacement changes caused by subtle movements through high-frequency sound waves. For example, it can identify whether the amplitude of the elderly's hand tremors exceeds the normal range through changes in ultrasonic signals, which can assist in evaluating the early symptoms of Parkinson's disease.

[0145] Among them, the surface motion characteristics refer to the motion form, motion trajectory and dynamic parameters that can be observed on the body surface of middle-aged and elderly people or captured by sensors.

[0146] During the specific implementation process, the ultrasonic frequency band can be used to identify the changes in the distance between key points on the body surface (such as joints and limbs) and the sensor, and to identify the gait, gestures and other body movement characteristics of middle-aged and elderly people.

[0147] The embodiment of the present invention obtains a multimodal voiceprint dataset by spatiotemporally fusing the internal organ vibration characteristics and the surface motion characteristics. This dataset can integrate physiological and motion information of different dimensions and construct a more comprehensive physical state representation system for middle-aged and elderly people. For example, the synchronous fusion of the infrasonic characteristics of heart vibration and the ultrasonic characteristics of arm waving can fully analyze the relationship between the cardiopulmonary function and limb coordination during exercise of the elderly.

[0148] As an embodiment of the present invention, the internal organ vibration features and the body surface motion features are spatially and temporally fused to obtain a multimodal voiceprint dataset, including:

[0149] Performing a time alignment operation on the internal organ vibration feature and the body surface motion feature to obtain a synchronized spatiotemporal feature;

[0150] Performing action mapping on the synchronous spatiotemporal features to obtain a synchronous feature matrix;

[0151] Performing feature fusion on the synchronized feature matrix to obtain a multimodal action vector;

[0152] Performing spatiotemporal encoding on the multimodal motion vector to obtain a spatiotemporal motion feature set;

[0153] The dataset of the spatiotemporal action feature set is reconstructed to obtain a multimodal voiceprint dataset.

[0154] Among them, the synchronous feature matrix refers to a two-dimensional matrix in which the time-aligned internal organ vibration features and surface motion features are arranged according to the time series dimension and the feature type dimension. The multimodal motion vector refers to a one-dimensional vector compressed or converted into features of different modes in the synchronous feature matrix through feature fusion technology (such as neural networks, feature splicing). The spatiotemporal motion feature set refers to a data set obtained after temporal modeling and spatial correlation analysis of the multimodal motion vector.

[0155] In the specific implementation process, based on the sensor synchronization trigger signal (such as GPS clock synchronization), the internal organ vibration characteristics (infrasonic time domain signals, such as heart vibration waveforms) and the surface motion characteristics (ultrasonic ranging sequences, such as gait timestamps) can be calibrated according to the acquisition timestamp to eliminate the millisecond time deviation and generate a synchronized spatiotemporal feature sequence; a physiological correlation mapping table of organ vibration and surface motion (such as the correlation coefficient between heart vibration frequency and walking cadence) is established, and the organ characteristics (such as 0.9Hz heart vibration frequency) and motion characteristics (such as 0.8 steps / second cadence) in the synchronized feature sequence are filled in according to the dimensions to construct a "organ-motion-time" three-dimensional synchronized feature matrix; a multi-stream convolutional neural network (Multi-Stream The infrasound frequency band features (such as frequency domain energy distribution) and ultrasonic frequency band features (such as motion velocity vector) in the synchronous feature matrix are cross-modally fused using CNN) or feature cascade methods to generate multimodal motion vectors containing spatiotemporal information; the Transformer encoder is used to perform temporal modeling on the multimodal motion vectors to capture the feature evolution in the time dimension (such as the coordinated changes in heart vibration amplitude and walking speed within 5 consecutive seconds), and at the same time, the spatial association between organs and actions is strengthened through the spatial attention mechanism to generate a spatiotemporal motion feature set; the spatiotemporal motion feature set is reorganized in the format of "sample ID-time segment-feature dimension", and labels are added (such as normal gait, abnormal breathing pattern) to generate a structured multimodal voiceprint dataset.

[0156] The embodiment of the present invention utilizes the multimodal voiceprint dataset to analyze the voiceprint feedback characteristics of the middle-aged and elderly people. By fusing the multi-dimensional acoustic information of organ vibration and body surface movement, it can accurately analyze the comprehensive characteristics of the body state and assist in judging health risks or abnormal behaviors.

[0157] Optionally, the voiceprint feedback feature can be trained using a machine learning model (such as a convolutional neural network) on the spatiotemporal motion feature set in the multimodal voiceprint dataset, extracting the frequency, timing and spatial correlation patterns of the fusion features of infrasound (organ vibration) and ultrasound (body surface movement), and identifying features such as the respiratory rhythm and gait stability of middle-aged and elderly people.

[0158] S5. Utilize the physiological feedback characteristics, the gas feedback characteristics, and the voiceprint feedback characteristics to conduct a behavior analysis on the middle-aged and elderly people and obtain a behavior monitoring report.

[0159] The embodiment of the present invention utilizes the physiological feedback characteristics, the gas feedback characteristics, and the voiceprint feedback characteristics to perform behavioral analysis on the middle-aged and elderly people, and obtains a behavioral monitoring report that can mine health risks and behavioral abnormalities from multi-dimensional physiological signals and behavioral data, thereby achieving accurate monitoring and early warning of the physical condition of the middle-aged and elderly people. For example, by analyzing abnormal heart rate, changes in respiratory gas composition, and fluctuations in voice fundamental frequency, abnormal behavioral tendencies such as dizziness and fatigue caused by hypoglycemia in the elderly can be discovered in a timely manner.

[0160] As an embodiment of the present invention, the physiological feedback characteristics, the gas feedback characteristics, and the voiceprint feedback characteristics are used to perform behavioral analysis on the middle-aged and elderly people to obtain a behavioral monitoring report, including:

[0161] Using the physiological feedback characteristics, assessing the physiological risk level of the middle-aged and elderly people;

[0162] Using the gas feedback characteristics, scoring the respiratory function of the middle-aged and elderly people to obtain a respiratory score;

[0163] Calculating the voiceprint health index of the middle-aged and elderly person using the voiceprint feedback feature;

[0164] Based on the physiological risk level, the breathing score and the voiceprint health index, the middle-aged and elderly people are warned of abnormal behavior events to obtain warning information;

[0165] Based on the early warning information, a behavioral recommendation plan for the middle-aged and elderly people is constructed to obtain a behavior monitoring report.

[0166] Among them, the physiological risk level refers to, the breathing score refers to, the voiceprint health index refers to, and the warning information refers to

[0167] During the specific implementation process, physiological feedback characteristics (such as heart rate, blood pressure and other data) can be input into a pre-trained Logistic regression model, combined with medical standard thresholds, to output low, medium and high physiological risk levels. For example, the real-time physiological data of middle-aged and elderly people (such as heart rate 55-85 beats / min, blood pressure 150 / 95mmHg, blood oxygen saturation 92%) are input into a trained Logistic model, combined with medical diagnostic thresholds such as hypertension and hypoxemia, and the output physiological risk level is "medium risk"; based on gas feedback characteristics (such as the eigenvector of the corrected gas spectrum), the support vector machine (SVM) is used to compare with the normal breathing gas feature library to calculate the similarity score as the breathing score; for sound The voiceprint feedback features detected in the infrasonic frequency band have a voice fundamental frequency jitter of 0.8%, and the cough frequency obtained in the ultrasonic frequency band is 4 times / hour. After the key features are screened by the random forest algorithm and normalized, the voiceprint health index (such as 82 / 100) is calculated; the physiological risk level is "medium risk and above", the respiratory score is "below 70 points", and the voiceprint health index is "below 75 points" as the warning threshold. Through the decision tree model, a comprehensive judgment is made. Because the respiratory score and the voiceprint health index do not meet the standards, the warning is triggered, and a warning message of "abnormal respiratory function and motor coordination, need attention" is generated; according to the warning information, the corresponding suggestions are retrieved from the health plan knowledge base to generate a behavior monitoring report, such as the recommendation for lung function testing.

[0168] Example 2:

[0169] like Figure 2 The figure shows the functional module diagram of the real-time monitoring and analysis system of the behavior data of the middle-aged and elderly people in the smart health care of the present invention.

[0170] The smart health care and elderly care behavior data real-time monitoring and analysis system 200 described in the present invention can be installed in an electronic device. Depending on the functions implemented, the smart health care and elderly care behavior data real-time monitoring and analysis system can include the signal acquisition module 201, physiological characteristics analysis module 202, gas characteristics analysis module 203, acoustic characteristics analysis module 204, and comprehensive characteristics analysis module 205. The modules described in the present invention, also known as units, refer to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function. They are stored in the memory of the electronic device.

[0171] In the embodiment of the present invention, the functions of each module / unit are as follows:

[0172] The signal acquisition module 201 is used to query the behavior time information of the middle-aged and elderly people in the smart health care environment, and based on the behavior time information, collect the bioresonance signal, gas spectrum and sound wave signal of the middle-aged and elderly people;

[0173] The physiological characteristic analysis module 202 is used to perform tensor field decomposition on the bioresonance signal to obtain a decomposed signal, construct a resonance characteristic spectrum of the middle-aged and elderly people using the decomposed signal, and analyze the physiological feedback characteristics of the middle-aged and elderly people using the resonance characteristic spectrum;

[0174] The gas characteristic analysis module 203 is configured to perform temperature compensation on the gas spectrum to obtain a corrected gas spectrum, perform feature compression on the corrected gas spectrum to obtain a gas characteristic vector, construct an odor-behavior association map of the middle-aged and elderly people using the gas characteristic vector, and identify the gas feedback characteristics of the middle-aged and elderly people using the odor-behavior association map;

[0175] The acoustic feature analysis module 204 is configured to perform frequency domain decomposition on the acoustic wave signal to obtain infrasonic frequency bands and ultrasonic frequency bands, identify the vibration characteristics of the internal organs of the middle-aged and elderly people using the infrasonic frequency bands, identify the surface motion characteristics of the middle-aged and elderly people using the ultrasonic frequency bands, perform spatiotemporal fusion of the internal organ vibration characteristics and the surface motion characteristics to obtain a multimodal voiceprint dataset, and analyze the voiceprint feedback characteristics of the middle-aged and elderly people using the multimodal voiceprint dataset;

[0176] The comprehensive feature analysis module 205 is used to perform behavior analysis on the middle-aged and elderly people by using the physiological feedback features, the gas feedback features and the voiceprint feedback features to obtain a behavior monitoring report.

[0177] In detail, the modules in the real-time monitoring and analysis system 200 for the behavior data of the elderly in the smart health care of the embodiment of the present invention are used in the same manner as above. Figure 1 The same technical means are used in the real-time monitoring and analysis of behavioral data of the elderly in smart health care, and can produce the same technical effects, so I will not go into details here.

[0178] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0179] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. 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 solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care, characterized by: The method comprises: Querying the behavior time information of the middle-aged and elderly people in the smart health care environment, and collecting the bioresonance signals, gas spectra and sound wave signals of the middle-aged and elderly people based on the behavior time information; performing tensor field decomposition on the bioresonance signal to obtain a decomposed signal, constructing a resonance characteristic spectrum of the middle-aged and elderly people using the decomposed signal, and analyzing the physiological feedback characteristics of the middle-aged and elderly people using the resonance characteristic spectrum; performing temperature compensation on the gas spectrum to obtain a corrected gas spectrum, performing feature compression on the corrected gas spectrum to obtain a gas feature vector, constructing an odor-behavior association map of the middle-aged and elderly people using the gas feature vector, and identifying the gas feedback characteristics of the middle-aged and elderly people using the odor-behavior association map; Decomposing the acoustic wave signal in the frequency domain to obtain an infrasonic frequency band and an ultrasonic frequency band, using the infrasonic frequency band to identify the vibration characteristics of the internal organs of the middle-aged and elderly people, and using the ultrasonic frequency band to identify the surface motion characteristics of the middle-aged and elderly people, spatiotemporally fusing the internal organ vibration characteristics and the surface motion characteristics to obtain a multimodal voiceprint dataset, and using the multimodal voiceprint dataset to analyze the voiceprint feedback characteristics of the middle-aged and elderly people; The physiological feedback characteristics, the gas feedback characteristics and the voiceprint feedback characteristics are used to perform behavior analysis on the middle-aged and elderly people to obtain a behavior monitoring report.

2. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1 is characterized in that: Based on the behavior time information, the bioresonance signal, gas spectrum and sound wave signal of the middle-aged and elderly person are collected, including: Based on the behavior time information, dividing the behavior time axis of the middle-aged and elderly people to obtain behavior time segments; Based on the behavior time segments, deploying trimodal sensors in the activity areas of the middle-aged and elderly people; The trimodal sensor is used to collect the biological resonance signals, gas spectra and sound wave signals of the middle-aged and elderly people.

3. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1 is characterized in that: Performing tensor field decomposition on the bioresonance signal to obtain a decomposed signal includes: performing quantum noise suppression on the bioresonance signal to obtain a noise-reduced three-dimensional tensor; Performing high-order tensor decomposition on the denoised three-dimensional tensor to obtain a core tensor and a factor matrix group; Performing physiological semantic mapping on the factor matrix group to obtain a physiological mapping tensor; performing joint tensor slicing processing on the core tensor and the physiological mapping tensor to obtain a physiological component tensor; The physiological component tensor is subjected to spatiotemporal reconstruction to obtain a decomposed signal.

4. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1, characterized in that: The physiological feedback characteristics of the middle-aged and elderly people are analyzed using the resonance characteristic spectrum, including: querying historical resonance characteristic data of the middle-aged and elderly people to construct a standard reference atlas of the middle-aged and elderly people; Extracting key characteristic parameters of the same latitude from the resonance characteristic spectrum and the standard reference spectrum to obtain comparison parameters; quantifying the difference of the comparison parameters to obtain a quantitative difference value; Based on the quantified difference value, abnormal marking is performed on the resonance characteristic spectrum to obtain a marked characteristic point; The physiological feedback characteristics of the middle-aged and elderly people are identified by using the marked feature points.

5. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1 is characterized in that: Using the gas feature vector, constructing the smell-behavior association map of the middle-aged and elderly people, including: Using the gas characteristic vector, constructing the behavior synchronization gas matrix of the middle-aged and elderly people; performing a graph node embedding operation on the behavior synchronization gas matrix to obtain an odor behavior node; Using the odor behavior nodes, constructing a behavioral dynamic graph of the middle-aged and elderly people; Performing heterogeneous graph convolution processing on the behavior dynamic graph to obtain behavior correlation features; The behavior association features are subjected to graph visualization processing to obtain an odor-behavior association graph.

6. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1, characterized in that: Using the odor-behavior association map, identifying the gas feedback characteristics of the middle-aged and elderly people includes: Performing key point screening on the odor-behavior association map to obtain a target behavior focus group; Based on the target behavior focus group, identifying the risk propagation link of the odor-behavior association graph; Utilizing the risk propagation link, the target focus area of ​​the middle-aged and elderly people is located to obtain feedback areas; Performing risk rating on the feedback parts to obtain a health risk level table; The health risk level table is used to identify the gas feedback characteristics of the middle-aged and elderly people.

7. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 6, characterized in that: Based on the target behavior focus group, identifying the risk propagation link of the odor-behavior association graph includes: Calculate the risk propagation probability of key points in the target behavior focus group: Among them, P i→j represents the risk propagation probability, i represents the key point i in the target behavior focus group, j represents the key point j in the target behavior focus group, i→j represents a link from node i to node j, W i，j represents the edge weight from key point i to j, N(i) represents the set of leading points of key point i, W i，k Represents the weight of key point i to adjacent point k; Based on the risk propagation probability, target links are filtered from the target behavior focus group to obtain risk propagation links.

8. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1, characterized in that: The internal organ vibration features and the body surface motion features are temporally and spatially fused to obtain a multimodal voiceprint dataset, including: Performing a time alignment operation on the internal organ vibration feature and the body surface motion feature to obtain a synchronized spatiotemporal feature; Performing action mapping on the synchronous spatiotemporal features to obtain a synchronous feature matrix; Performing feature fusion on the synchronized feature matrix to obtain a multimodal action vector; Performing spatiotemporal encoding on the multimodal motion vector to obtain a spatiotemporal motion feature set; The dataset of the spatiotemporal action feature set is reconstructed to obtain a multimodal voiceprint dataset.

9. The method for real-time monitoring and analysis of behavioral data of middle-aged and elderly people in smart health care according to claim 1, characterized in that: The physiological feedback characteristics, the gas feedback characteristics, and the voiceprint feedback characteristics are used to perform behavioral analysis on the middle-aged and elderly people to obtain a behavioral monitoring report, including: Using the physiological feedback characteristics, assessing the physiological risk level of the middle-aged and elderly people; Using the gas feedback characteristics, scoring the respiratory function of the middle-aged and elderly people to obtain a respiratory score; Calculating the voiceprint health index of the middle-aged and elderly person using the voiceprint feedback feature; Based on the physiological risk level, the breathing score and the voiceprint health index, the middle-aged and elderly people are warned of abnormal behavior events to obtain warning information; Based on the early warning information, a behavioral recommendation plan for the middle-aged and elderly people is constructed to obtain a behavior monitoring report.

10. The real-time monitoring and analysis system for the behavior data of middle-aged and elderly people in smart health care is characterized by: The system comprises: A signal acquisition module is used to query the behavior time information of the middle-aged and elderly people in the smart health care environment, and based on the behavior time information, collect the biological resonance signal, gas spectrum and sound wave signal of the middle-aged and elderly people; A signal acquisition module is used to query the behavior time information of the middle-aged and elderly people in the smart health care environment, and based on the behavior time information, collect the biological resonance signal, gas spectrum and sound wave signal of the middle-aged and elderly people; a physiological characteristic analysis module, configured to perform tensor field decomposition on the bioresonance signal to obtain a decomposed signal, construct a resonance characteristic spectrum of the middle-aged and elderly people using the decomposed signal, and analyze the physiological feedback characteristics of the middle-aged and elderly people using the resonance characteristic spectrum; a gas characteristic analysis module, configured to perform temperature compensation on the gas spectrum to obtain a corrected gas spectrum, perform feature compression on the corrected gas spectrum to obtain a gas characteristic vector, construct an odor-behavior association map of the middle-aged and elderly using the gas characteristic vector, and identify the gas feedback characteristics of the middle-aged and elderly using the odor-behavior association map; an acoustic feature analysis module for performing frequency domain decomposition of the acoustic wave signal to obtain infrasonic frequency bands and ultrasonic frequency bands, identifying the vibration characteristics of the internal organs of the middle-aged and elderly individuals using the infrasonic frequency bands, identifying the surface motion characteristics of the middle-aged and elderly individuals using the ultrasonic frequency bands, performing spatiotemporal fusion of the internal organ vibration characteristics and the surface motion characteristics to obtain a multimodal voiceprint dataset, and analyzing the voiceprint feedback characteristics of the middle-aged and elderly individuals using the multimodal voiceprint dataset; The comprehensive feature analysis module is used to use the physiological feedback features, the gas feedback features and the voiceprint feedback features to perform behavioral analysis on the middle-aged and elderly people and obtain a behavior monitoring report.

Images