Internet of Things-based Home Safety Monitoring System for Retired Elderly People

Abstract

The present invention belongs to the technical field of intelligent health monitoring. The present invention discloses a home safety monitoring system for retired elderly people based on the Internet of Things, including: deploying multiple types of sensors in the elderly people's home environment for all-weather data collection and analysis to obtain behavior data with activity feature tags; performing correlation analysis to obtain a personalized safety risk assessment model; performing real-time monitoring and pattern recognition to obtain a behavior trajectory analysis report; performing safety level classification and intelligent configuration to obtain a virtual safety guard area and an adaptive reminder rule; performing hierarchical classification processing and early warning signal generation to obtain hierarchical early warning information and response permission policies; performing integrated analysis to obtain a dynamic intervention strategy, and updating the personalized safety risk assessment model according to the dynamic intervention strategy to obtain a target safety monitoring model; effectively solving the blind spots and deficiencies of traditional manual monitoring and providing a reassuring home environment for the elderly.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent health monitoring, and more specifically, to a home safety monitoring system for retired elderly people based on the Internet of Things. Background Art

[0002] With the intensification of the aging population trend, the home safety problem of retired elderly people has become increasingly prominent. At the same time, the rapid development of Internet of Things technology has provided new possibilities for solving this problem. Traditional home monitoring solutions for the elderly mainly rely on simple call devices or passive camera monitoring, lacking intelligence and foresight, and unable to effectively prevent the occurrence of dangerous events such as falls, getting lost, or sudden illnesses. The highly individualized characteristics of the health status and behavior habits of the elderly population make the one-size-fits-all monitoring solution have limited effects and easily cause privacy and over-intervention problems.

[0003] The existing home safety monitoring technologies for the elderly mainly have the following problems: the perception means are single and lack semantic understanding ability, unable to accurately identify and analyze the complex behavior patterns of the elderly; the risk assessment model is too static and generalized, making it difficult to adapt to the dynamic changes of the health status and behavior habits of the elderly; the safety monitoring strategy for living spaces is rigid and cannot provide differential protection according to the risk characteristics of different regions; the early warning mechanism is not intelligent enough, either with too high sensitivity resulting in frequent false alarms, or being slow to respond resulting in missed reports of key risk events; the intervention response lacks systematicness and coordination, making it difficult to efficiently dispatch household and medical resources for timely rescue.

[0004] In view of this, the present invention proposes a home safety monitoring system for retired elderly people based on the Internet of Things to solve the above problems. Summary of the Invention

[0005] In order to overcome the above defects of the prior art and to achieve the above object, the present invention provides the following technical solutions:

[0006] Acquisition and Processing Module: Deploy multiple types of sensors in the home environment of the elderly for all-weather data acquisition and analysis to obtain behavior data with activity feature tags;

[0007] Model Construction Module: Conduct correlation analysis based on the behavior data and the elderly health record information to obtain a personalized safety risk assessment model;

[0008] Analysis Module: According to the personalized safety risk assessment model, conduct real-time monitoring and pattern recognition of the daily activities of the elderly to obtain a behavior trajectory analysis report;

[0009] Division Module: Based on the behavior trajectory analysis report, conduct safety level division and intelligent configuration of the living areas of the elderly to obtain virtual safety guard areas and adaptive reminder rules;

[0010] Hierarchical module: According to the activity feature tags, the virtual security guard area, and the adaptive reminder rules, perform hierarchical classification processing and early warning signal generation on the behavior data to obtain hierarchical early warning information and response permission policies;

[0011] Optimization module: Based on the behavior trajectory analysis report, the response permission policy, the processing records of the hierarchical early warning information, and external medical resource information, perform integrated analysis to obtain a dynamic intervention strategy, and update the personalized security risk assessment model according to the dynamic intervention strategy to obtain a target security guardianship model.

[0012] Furthermore, deploying multi-type sensors in the elderly's home environment for all-weather data collection and analysis to obtain behavior data with activity feature tags, including:

[0013] Conduct spatial planning and sensing requirement analysis on the elderly's home environment to obtain a sensor distribution plan, and deploy multi-type sensing devices based on the sensor distribution plan to construct an environment perception network;

[0014] Continuously collect environmental parameters and activity signals through the environment perception network to obtain an original perception data stream, and perform signal filtering and time series alignment processing on the original perception data stream to obtain cleaned structured data;

[0015] Perform spatio-temporal feature calculation and behavior sequence segmentation based on the structured data to obtain a basic behavior unit set, and apply pattern recognition algorithms and context correlation analysis to the basic behavior unit set to obtain behavior semantic features;

[0016] Perform activity classification and attribute annotation according to the behavior semantic features to obtain preliminary behavior labels, and perform verification and enhancement processing on the preliminary behavior labels in combination with the historical behavior pattern library to obtain behavior data with activity feature tags.

[0017] Furthermore, perform correlation analysis based on the behavior data and the elderly's health record information to obtain a personalized security risk assessment model, including:

[0018] Perform structured parsing and feature extraction on the elderly's health record information to obtain a health status feature vector, and identify health risk factors according to the health status feature vector to obtain an initial health risk rating;

[0019] Based on the initial health risk rating and the behavior data, perform correlation analysis on the health status-behavior pattern relationship to obtain a risk association rule set and a healthy behavior feature model, and perform cross-validation and weight assignment on the health status and behavior pattern according to the risk association rule set and the healthy behavior feature model to obtain a healthy behavior risk association map;

[0020] Calculate the importance of risk factors and prioritize them based on the health behavior risk association map to obtain a risk factor weight matrix, and quantify the levels of potential risk scenarios according to the risk factor weight matrix and the activity feature labels to obtain an initial risk assessment matrix;

[0021] Based on the initial risk assessment matrix, conduct a temporal pattern analysis of risk prediction to obtain a set of risk evolution rules, and based on the set of risk evolution rules, perform Markov chain modeling and risk transfer probability calculation on the initial risk assessment matrix to obtain a dynamic risk probability distribution;

[0022] Evaluate the risk prediction accuracy and optimize the threshold based on the dynamic risk probability distribution to obtain a set of risk model adjustment parameters, and based on the set of risk model adjustment parameters, conduct multi-feature fusion of risk assessment and model training verification to obtain a personalized security risk assessment model.

[0023] Furthermore, according to the personalized security risk assessment model, conduct real-time monitoring and pattern recognition of the daily activities of the elderly to obtain a behavioral trajectory analysis report, including:

[0024] Collect and process the activity data of the elderly in real time to obtain an original behavior data stream, and based on the original behavior data stream and the personalized security risk assessment model, define normal behaviors and detect abnormalities of the activities of the elderly to obtain an activity status marking result;

[0025] Based on the activity status marking result, extract the time-space characteristics of behavioral events and identify the activity types to obtain a multi-dimensional behavioral feature set, and based on the multi-dimensional behavioral feature set, analyze the activity patterns of the elderly and build a conventional behavior model to obtain a behavior pattern library;

[0026] Based on the behavior pattern library and historical activity data, construct a behavior time series model and conduct trend analysis to obtain a behavior change trajectory diagram, and based on the behavior change trajectory diagram and preset health and safety indicators, calculate behavior deviations and risk scores to obtain a behavior risk measurement table;

[0027] Based on the behavior risk measurement table and the personalized security risk assessment model, conduct risk level classification and urgency ranking to obtain a risk event priority queue, and through the risk event priority queue, conduct intelligent allocation and parallel monitoring of guardianship tasks to obtain a distributed guardianship task network;

[0028] Based on the distributed guardianship task network, conduct statistical aggregation and pattern refinement of behavior data to obtain a structured behavior record, and based on the structured behavior record, conduct activity correlation analysis and safety trend prediction to obtain a behavioral trajectory analysis report.

[0029] Further, based on the behavior trajectory analysis report, the living area of the elderly is divided into safety levels and intelligently configured to obtain a virtual safety protection area and an adaptive reminder rule, including:

[0030] Perform spatio-temporal distribution statistics and frequency analysis on the behavior trajectory analysis report to obtain a behavior pattern feature matrix, and perform living area correlation calculation and classification based on the behavior pattern feature matrix and environmental factor data to obtain an initial regional function set;

[0031] Based on the initial regional function set, perform heat map modeling and regional clustering analysis on the correlation between the activity trajectories of the elderly to obtain a behavior density distribution map, and perform safety risk assessment and level division based on the behavior density distribution map to obtain a preliminary safety area division result;

[0032] Based on the preliminary safety area division result, identify high-risk areas and analyze abnormal behaviors to obtain a risk area correlation matrix, and perform daily activity pattern modeling and scenario deduction based on the risk area correlation matrix to obtain a safety guardianship flow chart atlas;

[0033] Based on the safety guardianship flow chart atlas and the health status indicators of the elderly, set regional safety warning thresholds and deploy monitoring points to obtain a regional guardianship strategy set, and construct a virtual safety protection environment and configure sensors based on the regional guardianship strategy set to obtain a virtual safety protection area;

[0034] Based on the virtual safety protection area and the caregiver's permission settings, extract rules and parameterize the description of safety warning behaviors to obtain an adaptive reminder rule.

[0035] Further, according to the activity feature tags, the virtual safety protection area, and the adaptive reminder rule, perform hierarchical classification processing and warning signal generation on the behavior data to obtain hierarchical warning information and a response permission strategy, including:

[0036] Perform scenario analysis and feature extraction on the behavior data to obtain a set of behavior feature vectors, and perform abnormal behavior severity assessment and priority ranking based on the set of behavior feature vectors and the activity feature tags to obtain a hierarchical abnormal behavior task list;

[0037] Based on the hierarchical abnormal behavior task list and the virtual safety protection area, automatically select warning algorithms and configure parameters to obtain a differentiated warning plan, and perform parallel analysis and accuracy verification of abnormal behaviors according to the differentiated warning plan to obtain an initial warning data set;

[0038] Perform hierarchical organization and notification control marking of warning information based on the initial warning data set and the adaptive reminder rule to obtain a warning information packet with multi-level protection, and perform fine-grained division of response permissions and policy generation based on the warning information packet with multi-level protection to obtain an initial response permission rule set;

[0039] Perform modeling of the response process state machine and definition of conversion rules based on the initial response permission rule set and the monitoring scenario requirements to obtain a dynamic response state diagram, and perform response operation risk assessment and threshold setting based on the dynamic response state diagram to obtain response risk control parameters;

[0040] Generate and distribute push warning signals at different levels based on the response risk control parameters to obtain a distributed warning instruction set, and perform protocol design and effectiveness verification of the response process based on the distributed warning instruction set to obtain warning information and response permission policies at different levels.

[0041] Furthermore, integrate and analyze the behavior trajectory analysis report, the response permission policy, the processing records of the warning information at different levels, and external medical resource information to obtain a dynamic intervention strategy, including:

[0042] Perform time series correlation and pattern extraction on the behavior trajectory analysis report and the processing records of the warning information at different levels to obtain a health risk feature matrix, and perform abnormal state correlation calculation and classification based on the health risk feature matrix and the response permission policy to obtain an initial risk scenario set;

[0043] Based on the initial risk scenario set, perform trend graph modeling and type clustering analysis on the changing trend of the health status of the elderly to obtain a health status change graph, and perform medical needs assessment and urgency division based on the health status change graph to obtain a preliminary intervention needs division result;

[0044] Perform high-risk state identification and intervention effect analysis based on the preliminary intervention needs division result to obtain an intervention needs correlation matrix, and perform emergency response mode modeling and plan deduction based on the intervention needs correlation matrix to obtain an intervention flow chart atlas;

[0045] Set the threshold for medical resource allocation and plan the rescue path based on the intervention flow chart atlas and external medical resource information to obtain a resource allocation strategy set, and construct a dynamic intervention environment and configure a rescue team based on the resource allocation strategy set to obtain a dynamic intervention strategy.

[0046] Furthermore, update the personalized safety risk assessment model according to the dynamic intervention strategy to obtain a target safety monitoring model, including:

[0047] Associate and perform pattern mining on the dynamic intervention strategy and historical intervention records to obtain an intervention effect sequence diagram, and evaluate the effectiveness of the intervention measures and conduct adaptability detection based on the intervention effect sequence diagram and the changes in the health status of the elderly to obtain an intervention effect measurement scale;

[0048] Conduct risk factor correlation analysis based on the intervention effect measurement scale and the behavioral characteristics of the elderly to obtain a risk factor correlation network, and perform potential health risk reasoning and probability estimation based on the risk factor correlation network to obtain a health risk probability tree;

[0049] Optimize the allocation of guardianship resources and monitor the plan based on the health risk probability tree to obtain a preliminary guardianship strategy plan, and evaluate the feasibility of the guardianship measures and perform priority ranking based on the preliminary guardianship strategy plan to obtain a hierarchical guardianship strategy set;

[0050] Dynamically generate guardianship rules and perform consistency detection based on the hierarchical guardianship strategy set to obtain a target guardianship rule library, and automate the choreography of the daily guardianship process and construct a decision tree based on the target guardianship rule library to obtain an intelligent guardianship plan;

[0051] Dynamically adjust the risk assessment parameters and update the monitoring conditions based on the intelligent guardianship plan to obtain a temporary safety assessment strategy, and globally optimize and balance the adjustment of the personalized safety risk assessment model based on the temporary safety assessment strategy to obtain a target safety guardianship model.

[0052] The home safety guardianship device for retired elderly people based on the Internet of Things includes: a memory and at least one processor, and instructions are stored in the memory; when the at least one processor calls the instructions in the memory, the home safety guardianship device for retired elderly people based on the Internet of Things can execute the home safety guardianship system for retired elderly people based on the Internet of Things.

[0053] A computer-readable storage medium stores instructions thereon, and when it runs on a computer, it causes the computer to execute the home safety guardianship method for retired elderly people based on the Internet of Things.

[0054] The technical effects and advantages of the home safety guardianship system for retired elderly people based on the Internet of Things in the present invention:

[0055] By constructing a risk assessment model, the present invention achieves precise monitoring and avoids the one-size-fits-all problem of traditional monitoring systems. Through an environmental perception network constructed by multiple types of sensors, intelligent identification of abnormal behaviors and risk grading early warnings are realized, significantly improving the accuracy and timeliness of early warnings. Automatically divide the safe area based on the analysis of the behavior trajectory and perform intelligent configuration to form a virtual safe guard area, achieving key monitoring of high-risk areas. Self-optimize according to the intervention effect and changes in the health status, continuously update the risk assessment model, and improve the long-term monitoring effect. Combine external medical resource information to formulate dynamic intervention strategies to achieve efficient allocation of medical resources in case of emergencies. Adopt a hierarchical response permission strategy to protect the privacy and dignity of the elderly to the greatest extent while ensuring safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of a home safety monitoring system for retired elderly people based on the Internet of Things according to the present invention;

[0057] Figure 2 It is a schematic diagram of a home safety monitoring method for retired elderly people based on the Internet of Things according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0059] Embodiment 1;

[0060] Please refer to Figure 1 As shown, the home safety monitoring system for retired elderly people based on the Internet of Things in this embodiment includes:

[0061] Acquisition and processing module: Deploy multiple types of sensors in the elderly's home environment for all-weather data acquisition and analysis to obtain behavior data with activity feature tags;

[0062] Model construction module: Conduct correlation analysis based on the behavior data and the elderly's health record information to obtain a personalized safety risk assessment model;

[0063] Analysis module: According to the personalized safety risk assessment model, conduct real-time monitoring and pattern recognition of the elderly's daily activities to obtain a behavior trajectory analysis report;

[0064] Division module: Based on the behavior trajectory analysis report, conduct safety level division and intelligent configuration of the elderly's living area to obtain a virtual safety guard area and an adaptive reminder rule;

[0065] Classification module: Classify the behavior data according to the activity feature tags, the virtual safety protection area, and the adaptive reminder rules, and generate warning signals, so as to obtain classification warning information and response permission policies;

[0066] Optimization module: Integrate and analyze the behavior trajectory analysis report, the response permission policy, the processing records of the classification warning information, and the external medical resource information to obtain a dynamic intervention strategy, and update the personalized security risk assessment model according to the dynamic intervention strategy to obtain a target security monitoring model;

[0067] Each module is connected by wired and / or wireless means to achieve data transmission between modules.

[0068] Deploy multiple types of sensors in the elderly's home environment for all-weather data collection and analysis to obtain behavior data with activity feature tags;

[0069] Specifically, conduct spatial planning and sensing requirement analysis on the elderly's home environment. Through the division of functional areas in the living space and the study of the daily activity characteristics of the elderly, determine the key monitoring areas and monitoring parameters. The functional areas include main living spaces such as bedrooms, living rooms, kitchens, and bathrooms, as well as transition areas such as corridors and stairs. Based on the analysis results, formulate a sensor distribution plan to ensure that the sensor network covers all key areas, and configure appropriate types and quantities of sensing devices according to the characteristics and risk levels of different areas. Based on the sensor distribution plan, deploy multiple types of sensing devices in the home environment, including motion sensors, door and window sensors, pressure sensors, temperature and humidity sensors, smoke sensors, and smart cameras, etc., to build an all-round environmental perception network.

[0070] Continuously collect environmental parameters and activity signals through the environmental perception network to achieve all-weather monitoring of the elderly's living environment and behavior activities. The environmental parameters include environmental indicators such as temperature, humidity, light, and air quality, and the activity signals include behavior-related data such as motion detection, door and window status, and mattress pressure distribution. These data together constitute the original perception data stream, providing basic data support for subsequent behavior analysis. Perform signal filtering and time series alignment processing on the collected original perception data stream to eliminate sensor noise and interference factors during data collection, and synchronize and align the data of different sensors according to a unified time standard to obtain cleaned structured data, providing high-quality data basis for subsequent behavior analysis.

[0071] Perform spatio-temporal feature calculation and behavior sequence segmentation based on structured data. By analyzing time windows and extracting behavior features, identify the start and end points of the elderly's activities, and segment the continuous perceptual data stream into discrete behavior segments. These behavior segments form a set of basic behavior units, and each unit represents a relatively independent and complete behavior action or activity process. Apply deep learning pattern recognition algorithms and context correlation analysis to the set of basic behavior units to identify the internal features and semantic meanings of each behavior unit. At the same time, consider the environmental conditions, time background, and logical relationships of the preceding and following behaviors when occurring, and obtain behavior semantic features with a semantic understanding level.

[0072] Classify activities and annotate attributes according to the behavior semantic features. Classify the identified behavior patterns into daily activity types such as sleeping, eating, washing, resting, exercising, etc., and annotate key attributes such as duration, intensity, regularity, etc., to obtain preliminary behavior labels. To improve the label accuracy, the system will verify and enhance the preliminary behavior labels by combining with the historical behavior pattern library. Through comparison and similarity analysis with the elderly's past behavior patterns, correct possible recognition errors, and supplement detailed attributes related to personal habits. Finally, obtain accurate and comprehensive behavior data with activity feature labels.

[0073] In a specific embodiment, the deployment of multiple types of sensors in the elderly's home environment for all-weather data collection and analysis to obtain behavior data with activity feature labels may specifically include the following steps:

[0074] Conduct spatial planning and sensing requirement analysis on the elderly's home environment to obtain a sensor distribution plan, and deploy multiple types of sensing devices based on the sensor distribution plan to construct an environmental perception network;

[0075] Continuously collect environmental parameters and activity signals through the environmental perception network to obtain the original perceptual data stream, and perform signal filtering and time series alignment processing on the original perceptual data stream to obtain the cleaned structured data;

[0076] Perform spatio-temporal feature calculation and behavior sequence segmentation based on the structured data to obtain a set of basic behavior units, and apply pattern recognition algorithms and context correlation analysis to the set of basic behavior units to obtain behavior semantic features;

[0077] Classify activities and annotate attributes according to the behavior semantic features to obtain preliminary behavior labels, and verify and enhance the preliminary behavior labels by combining with the historical behavior pattern library to obtain behavior data with activity feature labels.

[0078] Specifically, conduct spatial planning and sensing requirement analysis for the home environment of the elderly. Using the systems engineering method, divide the home environment into main functional areas, such as bedrooms, living rooms, kitchens, bathrooms, corridors, and entrance areas. For each area, determine the monitoring requirement level through gerontological expert assessment and activity risk analysis. The monitoring requirement level can be expressed as a fusion function of functional importance, accident risk probability, and consequence severity: , where , and are weight coefficients, represents the monitoring requirement level, represents functional importance, represents accident risk probability, represents consequence severity, and satisfies the weight condition: . For example, due to the high risk of slipping in the bathroom, its accident risk probability and consequence severity are relatively high; the bedroom has a relatively high functional importance because it involves sleep monitoring. Based on the regional monitoring requirement level, formulate a sensor distribution plan, including sensor type selection, quantity determination, and installation location optimization. The sensor coverage density is proportional to the monitoring requirement level : , where is the proportionality coefficient. Based on the sensor distribution plan, deploy PIR motion sensors (bedrooms, living rooms, corridors), door magnetic sensors (entrances, kitchens, bathrooms), pressure sensors (mattresses, seats), temperature and humidity sensors (each room), smoke and gas sensors (kitchens), smart cameras (main activity areas, equipped with privacy protection technology), and water flow sensors (bathrooms, kitchens), etc. in the home environment. These sensors form an environmental perception network with a mesh topology through wireless communication protocols such as ZigBee, Z-Wave, or Wi-Fi, realizing seamless data collection and transmission.

[0079] Continuously collect environmental parameters and activity signals through the environmental perception network. The sampling frequency is dynamically adjusted according to the characteristics of parameter changes. For environmental parameters (temperature, humidity, light, etc.), a lower sampling rate is used (such as once every 5 minutes), and for activity signals (motion, pressure, etc.), a higher sampling rate is used (such as once every 10 seconds or event-triggered). The data collection process forms a multi-dimensional time series: , where represents the reading of the first sensor at time , represents the th sensor's reading at time . Perform signal filtering and time series alignment processing on the original perception data stream. Use the Kalman filter to eliminate random noise. The filtering process can be expressed as: , where is the estimated state at time , is the observed value at time , is the state transition matrix , is the Kalman gain. For outlier detection, the modified Z-score method is adopted: , is the normalized value represents the corrected data (such as environmental parameters), where and are the mean and standard deviation within the moving window respectively. When exceeds the preset threshold (usually set to 3.5), it is determined as an outlier and replaced or interpolated. Temporal alignment adopts the dynamic time warping (DTW) algorithm to unify the data of different sensors onto the normalized time axis to ensure the time consistency of the event sequence. After processing, the cleaned structured data is formed, where each structured data contains a timestamp, location information, and sensor readings.

[0080] Based on the structured data, spatio-temporal feature calculation and behavior sequence segmentation are carried out. The sliding time window method is adopted for preliminary segmentation, and the window size WID is dynamically adjusted according to the behavior type, with typical values ranging from 30 seconds to 5 minutes. Statistical features (mean, variance, peak value, etc.) and frequency domain features (the main frequency components are extracted through the fast Fourier transform) are calculated for the data within each window. The change point detection algorithm such as CUSUM (cumulative sum control chart) is used to identify the behavior conversion points. Through the identified behavior conversion points, the data stream is segmented into a series of basic behavior unit sets. The deep learning model is applied to the basic behavior unit sets for pattern recognition, and the long short-term memory network (LSTM) or temporal convolutional network (TCN) is used to process the temporal features. The model structure includes an input layer (feature dimension p), 1 to 2 hidden layers (each layer has 64 - 128 neurons), and an output layer (the number of behavior categories). The probability of each behavior pattern is calculated through the softmax function, and at the same time, context correlation analysis is carried out. The conditional random field (CRF) model is used to calculate the conditional probability of the behavior sequence. Through pattern recognition and context analysis, the behavior semantic features with rich semantic information are obtained.

[0081] Classify activities and annotate attributes according to behavioral semantic features, establish a behavioral ontology model, and map the identified behavioral patterns to a predefined activity category system. Activity categories include personal life activities (such as sleeping, eating, washing), instrumental daily activities (such as cooking, cleaning, going out), and social activities (such as making calls, receiving guests), etc. Annotate attributes for each activity, including basic attributes (duration, start and end time points, occurrence location) and extended attributes (activity intensity, completion quality, regularity). The activity intensity is calculated through the amplitude and change rate of relevant sensor signals: , where represents the activity intensity, is the activity index of the th signal, is the weight coefficient of the th signal. The completion quality is obtained through similarity evaluation with the standard pattern, and the value range is [0, 1]. These attribute assignments form the initial behavior label set T_initial. To improve the label accuracy, the system maintains a personalized historical behavior pattern library HIST, which contains the statistical characteristics and time distribution models of the elderly's past behavior patterns. Compare the initial labels with the historical patterns to obtain the label similarity. The comparison method uses a similarity comparison function, and common similarity comparison functions include cosine similarity. When the label similarity is lower than the preset similarity threshold, trigger the anomaly detection and label correction process, and use Bayesian inference or decision tree methods to optimize the labels. Finally, generate accurate and comprehensive behavioral data with activity feature labels, and each record contains the original data and rich behavioral label information.

[0082] Conduct correlation analysis based on the behavioral data and the elderly's health record information to obtain a personalized safety risk assessment model;

[0083] Specifically, conduct structured parsing and feature extraction on the elderly's health record information to obtain a health status feature vector, comprehensively capturing key health indicators such as the elderly's basic health status, chronic disease conditions, and physical function status. Apply natural language processing technology to the unstructured text data in the health record to extract key health descriptions and medical advice information, and at the same time standardize the structured data to unify the dimension and measurement standard. Convert the extracted health information into a numerical feature vector through feature engineering methods for subsequent analysis and processing.

[0084] Identify health risk factors based on the health status feature vector, analyze the feature vector using a medical knowledge graph and an expert rule system, identify health risk factors related to home safety, such as cardiovascular disease risk, cognitive decline risk, balance impairment risk, etc., and conduct a comprehensive assessment based on the identified risk factors to obtain an initial health risk rating.

[0085] Based on the initial health risk rating and behavioral data, a correlation analysis is conducted on the relationship between health status and behavioral patterns to explore the internal connection between health status and daily behavioral patterns. The correlation coefficient matrix between health characteristics and behavioral characteristics is calculated through statistical methods, significant relevant feature pairs are identified, and an association rule mining algorithm is applied to extract the association rules between health status and specific behavioral patterns, forming a risk association rule set. Meanwhile, a behavioral pattern prediction model based on health characteristics is constructed to capture the behavioral characteristic manifestations under different health statuses, obtaining a health behavior characteristic model.

[0086] According to the risk association rule set and the health behavior characteristic model, cross-validation and weight assignment are performed on health status and behavioral patterns, the reliability and importance of each risk association rule are evaluated, weight values are assigned to different rules, and a multi-layer association network is constructed between health status nodes and behavioral pattern nodes to form a health behavior risk association map, which can visually display the complex relationship between health status and behavioral risks.

[0087] Based on the health behavior risk association map, importance calculation and priority ranking of risk factors are carried out. The graph structure analysis method is applied to calculate the centrality index of each node in the map, the influence and importance of different risk factors in the overall risk network are evaluated, and combined with medical expert knowledge, the risk factors are ranked by importance to obtain a risk factor weight matrix. According to the risk factor weight matrix and activity characteristic labels, systematic evaluation and level quantification of potential risk scenarios in daily life are carried out, a risk scenario scoring system is established, and an initial risk assessment matrix is obtained.

[0088] Based on the initial risk assessment matrix, a time-series pattern analysis of risk prediction is conducted to study the evolution law of risk factors over time. Through time-series analysis methods, the periodic changes, trend changes, and mutation characteristics of risk indicators are mined, the time-series dependence relationship and triggering mechanism of risk indicators are identified, and a risk evolution law set is summarized. According to the risk evolution law set, a Markov chain model is built for the initial risk assessment matrix, a risk state transition probability matrix is constructed to describe the transition probability and evolution path between different risk states, and the steady-state distribution and expected duration of risk states are calculated to obtain a dynamic risk probability distribution.

[0089] Based on the dynamic risk probability distribution, conduct risk prediction accuracy evaluation and threshold optimization. Verify the accuracy and timeliness of the risk prediction model through historical data, evaluate the prediction accuracy of different risk types, and for the risk types with relatively large prediction errors, adjust the parameters of the prediction algorithm and the risk threshold settings to obtain a set of adjusted parameters for the risk model. Conduct multi-feature fusion of risk assessment based on the set of adjusted parameters for the risk model, integrate health features, behavior features, and environmental features, construct a multi-modal feature fusion framework, apply the ensemble learning method to train a personalized risk assessment model, evaluate the model performance through cross-validation and optimize the parameters, and finally obtain a personalized safety risk assessment model adapted to individual characteristics.

[0090] In a specific embodiment, the correlation analysis based on the behavior data and the elderly health record information to obtain a personalized safety risk assessment model may specifically include the following steps:

[0091] Conduct structured parsing and feature extraction on the elderly health record information to obtain a health status feature vector, and identify health risk factors based on the health status feature vector to obtain an initial health risk rating;

[0092] Based on the initial health risk rating and the behavior data, conduct a correlation analysis on the relationship between health status and behavior patterns to obtain a set of risk association rules and a health behavior feature model, and based on the set of risk association rules and the health behavior feature model, conduct cross-validation and weight assignment on the health status and behavior patterns to obtain a health behavior risk association map;

[0093] Based on the health behavior risk association map, calculate the importance and prioritize the risk factors to obtain a risk factor weight matrix, and based on the risk factor weight matrix and the activity feature labels, quantify the levels of potential risk scenarios to obtain an initial risk assessment matrix;

[0094] Based on the initial risk assessment matrix, conduct a temporal pattern analysis of risk prediction to obtain a set of risk evolution rules, and based on the set of risk evolution rules, conduct Markov chain modeling and risk transfer probability calculation on the initial risk assessment matrix to obtain a dynamic risk probability distribution;

[0095] Based on the dynamic risk probability distribution, conduct risk prediction accuracy evaluation and threshold optimization to obtain a set of adjusted parameters for the risk model, and based on the set of adjusted parameters for the risk model, conduct multi-feature fusion of risk assessment and model training and verification to obtain a personalized safety risk assessment model.

[0096] Specifically, for the structured analysis and feature extraction of the health record information of the elderly, natural language processing technology is used to process the text information such as diagnosis reports, medical orders, and medical histories in the health records. First, text tokenization and keyword extraction are performed. A dedicated tokenizer in the medical field is used to segment the text and extract medical terms and key descriptions. Then, named entity recognition technology is applied to identify medical entities such as disease names, symptom descriptions, and drug names, with an accuracy rate of over 95%. Finally, semantic analysis is used to determine the relationships between entities and construct a structured representation of the patient's health status. For structured data such as vital signs and laboratory test results, data cleaning and standardization processing are carried out to convert indicators with different dimensions into a unified standard score. The most representative health indicators are selected through a feature selection algorithm to construct a health status feature vector, and the vector dimension is usually 50 to 100, comprehensively describing the health status of the elderly.

[0097] Based on the health status feature vector, health risk factors are identified using an inference method based on a medical knowledge graph. The medical knowledge graph contains a triple relationship network of disease - symptom - risk. Through the graph matching algorithm, the health feature vector of the elderly is mapped into the knowledge graph to identify potential risk factors. The risk identification process can be expressed as: , where is the set of risk factors, represents the knowledge graph inference function, is the knowledge graph. For each identified risk factor, its risk intensity is calculated: , where represents the risk intensity, is the risk weight coefficient, is the indicator function, represents the indicator value, is the risk threshold. When the indicator value is greater than the risk threshold, it returns 1, and when the indicator value is less than the risk threshold, it returns 0. According to the set of risk factors and the corresponding intensities, a multi - level scoring model is applied to comprehensively evaluate the risk of the elderly and obtain an initial health risk rating. The rating is divided into five levels: extremely low risk (level 1), low risk (level 2), medium risk (level 3), high risk (level 4), and extremely high risk (level 5).

[0098] Based on the initial health risk rating and behavioral data, a correlation analysis of the health status - behavior pattern relationship is carried out. First, the Pearson correlation coefficient matrix between the health feature vector and the behavior feature vector is calculated. The calculation formula for each element in the correlation coefficient matrix is: , where, is the health feature, is the behavior feature, is the correlation calculation function. Common correlation calculation functions include cosine similarity. Select significant relevant feature pairs ( Greater than 0.6), as candidates for in-depth analysis. Then the Apriori algorithm is applied for association rule mining to calculate the support ( ) and confidence ( ): , , where represents the combination of health features, and represents the combination of behavior features. Select the rules where is greater than 0.1 and is greater than 0.7 to form the risk association rule set. At the same time, a healthy behavior feature model is constructed, and the random forest algorithm is used to establish the mapping relationship from health features to behavior features. The model consists of a forest of 500 decision trees, feature selection uses the Gini index, and training uses 10-fold cross-validation, and the model accuracy reaches more than 98%.

[0099] According to the risk association rule set and the healthy behavior feature model, cross-validation and weight assignment are performed on the health status and behavior patterns, and the lift ( ) of each association rule is calculated: , to evaluate the effectiveness of the rules. Sort the rules based on the lift value and assign weights: . Construct a healthy behavior risk association graph G=(V, E, W), where V is the set of nodes (including health status nodes and behavior pattern nodes), E is the set of association edges, and W is the set of edge weights. The graph adopts a multi-layer structure: the bottom-layer nodes represent specific health indicators and behavior features, the middle-layer nodes represent risk factors and behavior pattern categories, and the top-layer nodes represent comprehensive risk types. The connection weights between nodes are determined by the weights of the association rules, and the edge weights: . The graph is presented as an intuitive network structure through visualization technology, clearly showing the complex association relationship between health status and behavior risks.

[0100] Based on the healthy behavior risk association graph, calculate the importance and priority ranking of risk factors, calculate the centrality indicators of each node in the graph, including degree centrality, betweenness centrality, and eigenvector centrality. Combining these three centrality indicators, use the weight formula to calculate the comprehensive importance score of risk factors. Sort the risk factors according to the importance score to construct a risk factor weight matrix , where represents the impact weight of the risk factor on the risk scenario . Combining with the activity feature label set, calculate the risk score for each potential risk scenario: , where represents the activity feature in the scenario Salience in it. The scenarios are divided into different risk levels according to the risk scores, forming an initial risk assessment matrix. Each element in the initial risk assessment matrix represents the risk level of the elderly in the scenario.

[0101] Based on the initial risk assessment matrix, a time-series pattern analysis is carried out on the risk prediction, and a time series analysis method is used to process the continuously monitored risk index data. First, the Fourier transform is applied to identify the periodic pattern of the risk index: , where is the time function of the risk index, is the frequency domain representation, represents the angular frequency, represents the imaginary unit. Then, the ARIMA (Autoregressive Integrated Moving Average) model is used to capture the trend changes of the time series, and the change point detection algorithm is applied to identify the mutation points of the risk index. By analyzing these time series characteristics, a set of risk evolution rules is summarized to describe the evolution pattern of risk factors over time.

[0102] According to the set of risk evolution rules, a Markov chain modeling is carried out on the initial risk assessment matrix to construct a risk state transition probability matrix , and each element in the risk state transition probability matrix represents the probability of transferring from one risk state. The calculation method is: , is the risk state transferring to the risk state the number of observations, is the state the total number of observations. Based on the transition probability matrix, the steady-state distribution of the risk state is calculated, satisfying the distribution conditions: and , representing the probability distribution of the long-term risk level. Calculate the expected duration of the risk state: , where represents the expected duration of the risk state , indicating the average duration of a specific risk state. Combining the time series pattern and the Markov model, a dynamic risk probability distribution is generated to describe the probability distribution of each risk level at different time points.

[0103] Based on the dynamic risk probability distribution, the risk prediction accuracy is evaluated and the threshold is optimized, and the performance of the risk prediction model is verified using historical data. Calculate the prediction accuracy rate ( ), precision rate ( ), recall rate ( ) and the validation score . The calculation formula for the prediction accuracy rate is:

[0104] , and the calculation formula for the precision rate is: , the formula for recall rate is: ; the formula for verification score is: ;

[0105] where , , , represent true positive, true negative, false positive and false negative prediction results respectively. For the risk types with poor prediction performance, the risk threshold setting is optimized through ROC curve analysis, and the threshold point that maximizes the Youden index is selected. Adjust the parameters of the risk prediction algorithm, including the time window size, feature weights and model complexity, to form a set of risk model adjustment parameters.

[0106] Based on the set of risk model adjustment parameters, perform risk assessment multi-feature fusion and model training verification, and construct an ensemble learning framework to integrate multi-source feature information. Adopt a method that combines feature-level fusion and decision-level fusion: Feature-level fusion maps heterogeneous features to a unified feature space through feature transformation and standardization; Decision-level fusion integrates the prediction results of different models through multi-model voting or weighted average. The gradient boosting decision tree (GBDT) algorithm is used to train the personalized risk assessment model, and the model expression is:

[0107] , where is the basic decision tree, is the tree weight. The objective function is used for model training: , where is the mean squared error loss function, represents the actual item, represents the predicted item, and Ω is the regularization term. The performance of the model is evaluated through 5-fold cross-validation, and the hyperparameters, including tree depth, learning rate and regularization strength, are optimized using the grid search method. Finally, a personalized safety risk assessment model is obtained, which can evaluate potential safety risks in real time according to the health status, behavior patterns and environmental factors of the elderly, and provide a scientific basis for preventive measures.

[0108] According to the personalized safety risk assessment model, the daily activities of the elderly are monitored in real time and pattern recognition is performed to obtain a behavior trajectory analysis report;

[0109] Specifically, real-time collection and signal processing of the activity data of the elderly are carried out to obtain the original behavior data stream. A distributed sensor network is used for multi-source data collection, including motion sensors, pressure sensors, sound sensors, wearable devices, etc., to comprehensively capture the activity information of the elderly. The collected original signals are subjected to denoising, filtering, and feature enhancement processing to eliminate environmental interference and equipment errors and improve data reliability. Through signal processing, the heterogeneous data of different sensors are converted into a time-series data stream in a unified format, providing a basis for subsequent analysis. Based on the original behavior data stream and the personalized safety risk assessment model, the normal behavior of the elderly's activities is defined and abnormal detection is carried out to obtain the activity status marking result.

[0110] Based on the activity status marking result, time-space feature extraction and activity type recognition are carried out on the behavior events to obtain a multi-dimensional behavior feature set. By applying feature engineering methods to the marked behavior events, time features (such as duration, occurrence frequency, time distribution) and space features (such as activity area, movement trajectory, position change) are extracted to construct a multi-dimensional feature vector. A hierarchical classification model is used to identify the activity type of the behavior events, distinguishing different activity categories such as daily life activities like eating, sleeping, washing, watching TV, etc. Based on the multi-dimensional behavior feature set, the activity pattern of the elderly is analyzed and a conventional behavior model is built to obtain a behavior pattern library.

[0111] Based on the behavior pattern library and historical activity data, a behavior time-series model is constructed and trend analysis is carried out to obtain a behavior change trajectory graph. Through time-series analysis methods, a behavior time-series prediction model is established to capture the seasonal changes, periodic patterns, and long-term trends of the behavior pattern. Combining historical data, the change rate and deviation degree of the behavior indicators are calculated to generate a trajectory graph reflecting the behavior evolution process. Based on the behavior change trajectory graph and the preset health and safety indicators, behavior deviation calculation and risk scoring are carried out to obtain a behavior risk measurement table.

[0112] Based on the behavior risk measurement table and the personalized safety risk assessment model, risk level classification and urgency ranking are carried out to obtain a risk event priority queue. Through hierarchical analysis and risk weight calculation, risk level classification and priority ranking are carried out on the detected abnormal events to ensure that high-risk events can be responded to in a timely manner. Through the risk event priority queue, intelligent allocation and parallel monitoring of guardianship tasks are carried out to obtain a distributed guardianship task network.

[0113] Based on the distributed guardianship task network, statistical aggregation and pattern refinement of the behavior data are carried out to obtain a structured behavior record. Through data aggregation algorithms, the behavior data collected by distributed monitoring nodes are summarized and statistically analyzed to extract key behavior features and pattern features. Based on the structured behavior record, activity correlation analysis and safety trend prediction are carried out to obtain a behavior trajectory analysis report.

[0114] In a specific embodiment, the real-time monitoring and pattern recognition of the daily activities of the elderly according to the personalized security risk assessment model to obtain a behavioral trajectory analysis report may specifically include the following steps:

[0115] Real-time collect and process the activity data of the elderly to obtain the original behavior data stream, and based on the original behavior data stream and the personalized security risk assessment model, define the normal behavior and detect abnormalities of the activities of the elderly to obtain the activity status marking result;

[0116] Based on the activity status marking result, extract the time-space characteristics and identify the activity types of the behavioral events to obtain a multi-dimensional behavioral feature set, and based on the multi-dimensional behavioral feature set, analyze the activity pattern of the elderly and build a conventional behavior model to obtain a behavior pattern library;

[0117] Based on the behavior pattern library and historical activity data, construct a behavior time series model and perform trend analysis to obtain a behavior change trajectory graph, and based on the behavior change trajectory graph and the preset health and safety indicators, calculate the behavior deviation and risk score to obtain a behavior risk measurement table;

[0118] Based on the behavior risk measurement table and the personalized security risk assessment model, perform risk level classification and urgency ranking to obtain a risk event priority queue, and through the risk event priority queue, perform intelligent allocation and parallel monitoring of the guardianship tasks to obtain a distributed guardianship task network;

[0119] Based on the distributed guardianship task network, perform statistical aggregation and pattern refinement of the behavior data to obtain a structured behavior record, and based on the structured behavior record, perform activity correlation analysis and safety trend prediction to obtain a behavioral trajectory analysis report.

[0120] Specifically, for the real-time collection and signal processing of the activity data of the elderly, an Internet of Things sensing system composed of a PIR motion sensor, a pressure sensor, a smart bracelet, etc. is used to collect the activity data of the elderly at a sampling frequency of 5Hz - 20Hz. The PIR sensor detects the human activities in the room, the pressure sensor monitors the occupancy of positions such as the bed and the sofa, and the smart bracelet records physiological parameters such as heart rate, walking steps, and wrist activities. Signal processing is performed on the collected original signals, including median filtering to remove spike noise, wavelet transform to eliminate baseline drift, and Kalman filtering to fuse multi-source data. The signal processing process can be expressed as: , where is the original signal, represents the combination of filtering functions, is the processed signal. The processed data of each sensor is aligned according to the time stamp to form a unified original behavior data stream , where each data point contains the time stamp and sensor readings. Based on the original behavior data stream and the personalized security risk assessment model, normal behavior is defined and abnormal detection is carried out, adopting an adaptive threshold method based on individual historical behavior. Define the normal behavior boundary , where is the mean of historical behavior at a specific time point, is the standard deviation of historical behavior at a specific time point, is an adjustable parameter (usually taking 2 - 3). When the current behavior index exceeds the boundary , it is marked as a potential anomaly. For potential anomalies, verification based on spatio - temporal context is adopted, and the anomaly score is calculated:

[0121] , where, is the anomaly score, is the time deviation component, is the space deviation component, is the context deviation component, 、 and are weight coefficients and satisfy the weight condition: . When exceeds the preset anomaly threshold, it is confirmed as an abnormal behavior. Through this process, the activity status marking result is obtained:

[0122] , where is the timestamp, indicates the normal or abnormal status.

[0123] Based on the activity status marking result, spatio - temporal features of behavior events are extracted. The sliding window method is used to divide the continuous data stream into event segments of a fixed length. For each event segment, the time feature vector is extracted, where represents the duration, represents the start time, represents the occurrence frequency within 24 hours, represents the coefficient of variation of time regularity. At the same time, the space feature vector is extracted, where represents the activity location encoding, represents the activity coverage area, represents the length of the movement path, Denote the moving speed. Apply principal component analysis (PCA) to the extracted spatio-temporal features for dimensionality reduction, obtaining the dimensionality-reduced spatio-temporal feature vector. Use an ensemble classifier composed of support vector machine (SVM) and random forest (RF) for activity type recognition. Define the set of activity categories, including common daily activities such as getting up, washing, eating, watching TV, taking medicine, etc. The ensemble classifier outputs a probability vector, representing the probabilities that the event belongs to each activity category. The final recognition result is the category corresponding to the highest probability. Through the time features, space features, and activity types, a multi-dimensional behavior feature set is constructed. Based on the multi-dimensional behavior feature set, analyze the activity patterns of the elderly and conduct conventional behavior modeling, using Gaussian mixture model (GMM) to represent the multi-modal distribution characteristics of behaviors. For each type of activity, establish a GMM model. The GMM parameters are learned from historical data through the expectation-maximization algorithm. The set of GMM models for all activity categories constitutes the behavior pattern library.

[0124] Construct a behavior time series model based on the behavior pattern library and historical activity data, using long short-term memory network (LSTM) to capture the time dependence of the behavior sequence. The input of the LSTM model is the feature vector at time t and the hidden state at the previous time, and the output is the current hidden state and the predicted value. Through the trained LSTM model, predict the future behavior sequence, compare it with the actual observed value, and calculate the change trend of the behavior trajectory. Behavior change trajectory diagram: Denote at different time points of the behavior deviation degree . Calculate the behavior deviation according to the behavior change trajectory diagram and the preset health and safety indicators, and define the behavior deviation index , where is the actually observed behavior index, is the expected normal value, is the standard deviation of this index, is the weight coefficient, is the rd index category. The health and safety indicators include activity frequency indicators, diet regularity indicators, sleep quality indicators, medication compliance indicators, etc. Calculate the risk score based on the deviation index , where is the adjustment coefficient to control the sensitivity of the score. The risk score ranges from 0 to 100, and the higher the score, the greater the risk. Generate a behavior risk measurement table according to the risk scores of multiple indicators:

[0125] , where represents the index name, represents the corresponding risk score.

[0126] The risk level is divided based on the behavioral risk measurement scale and the personalized security risk assessment model, and the fuzzy logic method is used to handle the uncertainty in risk assessment. Define the risk level set, including: low risk, medium risk, high risk, and emergency risk. The corresponding membership function represents the degree to which the risk score belongs to the risk level. Calculate the priority score according to the risk level and the degree of urgency:

[0127] , where RL is the risk level value (1 - 4), U is the degree of urgency (time sensitivity), is the intervention difficulty, 、 and are the weight coefficients. Sort the risk events according to the priority score to form a risk event priority queue:

[0128] , where . Intelligently allocate the guardianship tasks through the risk event priority queue, and adopt a task allocation algorithm based on resource constraints. Define the guardianship resource set, including family caregivers, community service personnel, remote medical resources, etc. The resource allocation model considers the resource capacity , resource availability and the task-resource matching degree , and the objective function is to maximize the risk response efficiency: , where ∈{0,1} indicates whether to assign the event to the resource , and satisfy the resource constraint . Through the parallel monitoring mechanism, construct a distributed guardianship task network , where represents the optimal resource assigned to the event .

[0129] Statistically aggregate the behavior data based on the distributed guardianship task network, and adopt the hierarchical time aggregation method to summarize the behavior data from different time scales such as hours, days, weeks, and months. Define the aggregation function , where is the original data set, is the time point, This is the aggregation level. Through aggregation functions, statistical features such as mean, median, variance, trend, etc. are extracted to form structured behavior records. Activity correlation analysis is carried out based on the structured behavior records, and association rule mining algorithms are applied to discover the association relationships between behavior patterns. Define the association rule rule: X→Y, where X and Y are sets of behavior events, support supp(r)=P(X∪Y) represents the probability that X and Y occur simultaneously, and confidence conf(r)=P(Y|X) represents the probability that Y occurs under the condition that X occurs. Extract significant association rules with support and confidence exceeding the thresholds to form a behavior association network. Based on historical data and the discovered behavior patterns, a time series prediction model is applied for security trend prediction. An ARIMA(p,d,q) model is constructed for key security indicators, where p is the number of autoregressive terms, d is the order of differencing, and q is the number of moving average terms. The model is used to predict the change trends of security indicators in the next 7 days and 30 days to identify potential risk increase points. Integrate the results of activity correlation analysis and security trend prediction to generate a behavior trajectory analysis report, including content such as a summary of behavior changes, risk assessment results, early warning prompts, and intervention suggestions.

[0130] Based on the behavior trajectory analysis report, conduct safety level division and intelligent configuration for the living areas of the elderly to obtain virtual safety guard areas and adaptive reminder rules;

[0131] Specifically, conduct spatio-temporal distribution statistics and frequency analysis on the behavior trajectory analysis report, quantitatively statistically analyze the activities of the elderly in different time periods and different spatial positions, and extract time regularities and spatial distribution characteristics. Through time series analysis of the daily behaviors of the elderly, identify regular activity patterns and irregular activity patterns, and calculate the occurrence frequency, duration, and time distribution characteristics of various behaviors. At the same time, conduct density analysis and path tracking on spatial activities to identify high-frequency activity areas and regular movement paths. Integrate these spatio-temporal distribution and frequency data into a behavior pattern feature matrix, which comprehensively describes the behavior habits and living rules of the elderly and provides a data basis for subsequent security analysis.

[0132] Calculate and classify the correlation of the living areas according to the behavior pattern feature matrix and environmental factor data. The environmental factor data includes environmental parameters such as indoor lighting conditions, temperature and humidity changes, noise levels, and floor materials that may affect the safety of the elderly. By calculating the correlation coefficient between behavior characteristics and environmental factors, identify the influence degree and action mode of environmental factors on the behavior of the elderly. Based on the results of the correlation analysis, conduct functional classification of the living areas, divide them into different functional areas such as bedroom rest areas, kitchen activity areas, bathroom areas, living room communication areas, etc., to form an initial regional function set. This functional classification is not only based on the physical layout of the room, but also takes into account the actual usage methods and behavior characteristics of the elderly, and is more in line with personalized living habits.

[0133] Based on the initial regional function set, a heat map modeling and regional clustering analysis are carried out on the correlation between the activity trajectories of the elderly. By mapping the elderly activity data into a spatial coordinate system and using the activity intensity as the heat value, an intuitive activity heat map is generated. The high-temperature areas in the heat map represent areas with frequent activities, the low-temperature areas represent areas with less activities, and the temperature gradient reflects the gradual change relationship of activity frequencies. Based on the heat map, a density clustering algorithm is used to conduct a clustering analysis on the activity areas, identify area clusters with similar activity characteristics, and calculate the center points, ranges, and density characteristics of each cluster to form a behavior density distribution map. This distribution map not only shows the concentration degree of spatial activities but also reflects the activity connections and conversion frequencies between different areas.

[0134] Based on the behavior density distribution map, a safety risk assessment and grading are carried out. According to the activity density, environmental factors, and the health status of the elderly, the potential safety risks of each area are evaluated. The risk assessment takes into account various safety factors such as fall risk, electrical safety, scald risk, slip risk, etc., and conducts a comprehensive assessment in combination with the physical condition and cognitive ability of the elderly. According to the risk assessment results, the living areas are divided into three grades: high-risk areas, medium-risk areas, and low-risk areas, forming a preliminary safety area division result. High-risk areas usually include areas with relatively large potential hazards such as bathrooms and kitchens; medium-risk areas include areas where certain risks may exist such as stairs and corridors; low-risk areas are the areas where the elderly rest and carry out safe activities daily.

[0135] Based on the preliminary safety area division result, high-risk area identification and abnormal behavior analysis are carried out. The characteristics of high-risk areas are analyzed in depth to identify specific risk factors and possible dangerous situations. At the same time, the abnormal behavior patterns of the elderly in high-risk areas are analyzed, such as behavior characteristics that may indicate safety problems, such as staying in the bathroom for too long, getting up frequently at night, and irregular activities in the kitchen area. The correlation between the risk area characteristics and the abnormal behavior patterns is quantified into a risk area association matrix, which describes the association strength between specific areas, specific behaviors, and safety risks, providing an accurate positioning basis for safety guardianship.

[0136] According to the risk area association matrix, a daily activity pattern modeling and scenario deduction are carried out. A state transition model of the elderly's daily activities is constructed to describe the conversion rules and conditions between different activity states. Based on this model, a variety of safety risk scenarios are deduced and predicted, simulating the dangerous situations that may occur under different conditions and corresponding countermeasures. Through scenario deduction, key monitoring points and decision nodes are identified, and a complete safety guardianship flow chart is constructed. This flow chart describes the complete guardianship process from risk monitoring, abnormal identification to early warning response in the form of a directed graph, clarifying the triggering conditions and processing logics of each link.

[0137] Set regional safety warning thresholds and deploy monitoring points based on safety guardianship flowcharts and elderly health status indicators. Set personalized warning thresholds for different safety parameters according to the health indicators of the elderly, such as age, physical condition, and medical history. For example, for the elderly with mobility difficulties, lower the sensitivity threshold for fall detection; for patients with heart disease, increase the monitoring frequency of abnormal heart rates. On the basis of determining the warning thresholds, scientifically plan the deployment locations and densities of sensors to form a comprehensive and focused monitoring network, and finally form a set of regional guardianship strategies. This set of strategies includes the monitoring parameters, warning thresholds, response strategies, and upgrade mechanisms for each region, providing a decision-making basis for the safety protection system.

[0138] Build a virtual safety protection environment and configure sensors based on the set of regional guardianship strategies. Map the physical space into a virtual guardianship environment to achieve real-time status monitoring and risk warning. In the virtual environment, deploy various sensors according to the guardianship strategies, including human activity sensors, environmental parameter sensors, image recognition devices, etc., to build a multi-dimensional perception network. At the same time, establish a fusion processing mechanism for sensor data to improve the accuracy and reliability of risk recognition through comprehensive analysis of multi-source data. The finally formed virtual safety protection area is an organic combination of the physical space and digital guardianship, maintaining the natural comfort of the living environment while achieving unperceived safety protection.

[0139] Based on the virtual safety protection area and the caregiver's permission settings, extract rules and parameterize the description of safety warning behaviors, and analyze the best response strategies and notification mechanisms in different safety risk situations. According to the risk level and urgency, formulate hierarchical response rules to clarify in what situations to notify the family members, in what situations to automatically alarm, and in what situations the system processes autonomously. At the same time, considering the different roles and permissions of caregivers, design personalized notification content and methods to ensure the accuracy and timeliness of information transmission. The finally formed adaptive reminder rules can dynamically adjust the response strategies according to the actual situation, avoiding over-interference while ensuring safety, and balancing the relationship between safety protection and life privacy.

[0140] In a specific embodiment, the safety level division and intelligent configuration of the elderly living area based on the behavior trajectory analysis report to obtain the virtual safety protection area and the adaptive reminder rules may specifically include the following steps:

[0141] Conduct spatio-temporal distribution statistics and frequency analysis on the behavior trajectory analysis report to obtain a behavior pattern feature matrix, and calculate and classify the correlation of the living area according to the behavior pattern feature matrix and environmental factor data to obtain an initial regional function set;

[0142] Based on the initial regional function set, perform heat map modeling and regional clustering analysis on the correlation between the activity trajectories of the elderly to obtain a behavior density distribution map, and perform safety risk assessment and grading based on the behavior density distribution map to obtain a preliminary safety area division result;

[0143] Based on the preliminary safety area division result, identify high-risk areas and analyze abnormal behaviors to obtain a risk area correlation matrix, and perform daily activity pattern modeling and scenario deduction based on the risk area correlation matrix to obtain a safety guardianship flow chart atlas;

[0144] Based on the safety guardianship flow chart atlas and the health status indicators of the elderly, set regional safety warning thresholds and deploy monitoring points to obtain a regional guardianship strategy set, and construct a virtual safety guardianship environment and configure sensors based on the regional guardianship strategy set to obtain a virtual safety guardianship area;

[0145] Based on the virtual safety guardianship area and the caregiver's permission settings, extract rules and parametric descriptions for safety warning behaviors to obtain adaptive reminder rules.

[0146] Specifically, conduct spatio-temporal distribution statistics and frequency analysis on the behavior trajectory analysis report, using time series analysis and spatial statistics methods. In the time dimension, divide 24 hours into multiple time periods (such as one period every 2 hours), and count the activity frequency and type of the elderly in each time period. Assume represents the activity type in the time period The occurrence frequency, then the time distribution characteristic can be expressed as the time series { }. In the spatial dimension, divide the living space into grid cells, and calculate the residence time ratio and activity intensity of the elderly in each grid cell. Assume represents the coordinate The activity intensity at the location, then a spatial distribution characteristic map can be constructed. Combine the time characteristics and spatial characteristics to form a behavior pattern characteristic matrix where the matrix element represents the activity type in the time period at the location The occurrence intensity. Calculate the correlation between the behavior intensity and environmental factor data, and the environmental factor data includes illumination brightness, temperature, humidity, and noise level. Calculate the correlation coefficient where Represents the environmental factor vector, which consists of lighting brightness, temperature, humidity, and noise level. The correlation coefficient is calculated through a similarity measurement formula, such as the cosine similarity measurement formula. Based on the results of the correlation analysis, the K-means or fuzzy C-means clustering algorithm is used to classify the living areas by function, obtaining the initial regional function set, and each element in the set represents a functional area.

[0147] Based on the initial regional function set, a heat map model is built for the correlation between the activity trajectories of the elderly. The activity heat map The calculation formula is: , where represents the activity intensity of the th type of activity at location , represents the weight coefficient of the activity type, reflecting the importance of this activity to safety. After the heat map is generated, the DBSCAN density clustering algorithm is used for regional clustering analysis. The DBSCAN algorithm is based on two parameters: the neighborhood radius and the minimum number of points, and identifies the high-density areas in the heat map as activity clusters. For a point Point, if the number of points within its neighborhood radius is greater than or equal to the minimum number of points, then Point is a core point; if a point Po is within the neighborhood radius of the core point Point, then Point and Po belong to the same cluster. In this way, the natural clustering of the elderly's activities is identified, forming a behavior density distribution map , where represents the activity cluster type and density value to which the location belongs. Based on the behavior density distribution map, a safety risk assessment is carried out, and the risk score is obtained by weighted summation of the activity density, environmental risk factors, and personal risk factors (such as age, physical condition, etc.). According to the risk score, the area is divided into high-risk areas ( ), medium-risk areas ( ), and low-risk areas ( ) based on the preset regional risk thresholds, obtaining the preliminary safety area division result, where represents the risk score, represents the high-risk threshold, represents the medium-risk threshold.

[0148] Based on the preliminary safety area division result, high-risk area identification and abnormal behavior analysis are carried out, and the behavior patterns in the high-risk areas are analyzed in depth. Define the abnormal behavior index , indicating the degree of abnormality of the activity at time at location :

[0149] , where represents the historical average activity intensity, represents the standard deviation. If the value is greater than or equal to a preset activity anomaly threshold, it is determined as abnormal behavior. Based on the analysis results of high-risk areas and abnormal behaviors, a risk area association matrix is constructed. The matrix element represents the risk association intensity between area and area . The calculation formula is: , where represents the frequency of activity transfer from area to area , and respectively represent the risk scores of areas and . Based on the risk area association matrix, a daily activity pattern model is built, and the Hidden Markov Model (HMM) is used to describe the activity state transition of the elderly. The HMM model includes a state set, an observation set, a state transition probability matrix, an observation probability matrix, and an initial state distribution. Among them, the state set corresponds to different activity states, and the observation set corresponds to the observation results of sensors. The HMM parameters are trained through the Baum-Welch algorithm to obtain a state transition model that can describe the daily activity rules of the elderly. Based on the HMM model, a scenario deduction of the Monte-Carlo method is carried out to simulate and generate possible activity sequences and corresponding risk situations, forming a safety guardianship flow chart spectrum. The safety guardianship flow chart spectrum is a directed graph, where the nodes represent activity states, the edges represent state transitions, the node attributes include risk levels and monitoring strategies, and the edge attributes include transition probabilities and trigger conditions.

[0150] Based on the safety guardianship flow chart spectrum and the health status indicators of the elderly, the regional safety warning threshold is set. The health status indicators include an age index, a physical function index, and a disease risk index. The safety warning threshold is calculated using a weighted formula based on the health status indicators. According to the set safety warning threshold, the deployment positions and densities of monitoring points are planned. For area , the monitoring point density is calculated by the formula: , where represents the basic monitoring density, represents the risk score of area , represents the risk sensitivity coefficient. The specific positions of the monitoring points are determined by the maximum coverage algorithm, and the goal is to use the fewest monitoring points to achieve the maximum coverage of key areas. Combining the warning threshold and the monitoring point deployment plan, a regional guardianship strategy set is formed, where represents the area, represents the monitoring parameters, represents the warning threshold, represents the response strategy. Based on the regional guardianship strategy set, a virtual security guard environment is constructed. The virtual environment adopts three-dimensional modeling and digital twin technology to map the physical space into a visual monitoring model. In the virtual environment, a sensor network is configured according to the guardianship strategy set, including motion sensors, pressure sensors, temperature and humidity sensors, cameras, etc. The data of various sensors are preliminarily processed by edge computing nodes and then uploaded to the cloud for comprehensive analysis. Through data fusion algorithms, multi-source sensor data are integrated to improve the accuracy of event recognition, and finally a virtual security guard area is formed.

[0151] Based on the virtual security guard area and the caregiver's permission settings, rule extraction and parametric description are performed on the security warning behavior. The warning rules consist of trigger conditions, response strategies, and notification objects. The trigger condition is a logical expression based on sensor data and thresholds, such as "if (stay time is greater than the stay threshold) AND (location ∈ high-risk area)". The response strategy is determined according to the risk level and urgency, including local reminders, remote notifications, automatic intervention levels, etc. The notification object is determined according to the caregiver's permission and the relationship with the elderly to ensure that information is transmitted to the appropriate caregiver. Through machine learning methods, warning rule patterns are extracted from historical data and parametrically described. The validity score of the rule is calculated by the following formula: , where represents accurate warnings, represents false alarms, represents missed alarms. Based on the scoring results, the rules are optimized and adjusted, and finally an adaptive reminder rule set is formed. The adaptive reminder rule can dynamically adjust parameters according to the behavior changes of the elderly and environmental conditions to achieve personalized security protection.

[0152] According to the activity feature label, the virtual security guard area, and the adaptive reminder rule, the behavior data is classified by level and warning signals are generated to obtain hierarchical warning information and response permission policies;

[0153] Specifically, scene analysis and feature extraction are performed on the behavior data to establish a multi-dimensional feature vector including behavior, location, time, and environmental factors, comprehensively capturing the context information and feature attributes of the elderly's abnormal behavior. Based on the extracted feature vector and the pre-labeled activity feature label, the severity of the abnormal behavior is evaluated using the multi-index hierarchical analysis method, a priority ranking model is constructed, and a list of abnormal behavior tasks classified by urgency is generated. Through this list, the system can reasonably allocate computing resources and give priority to processing high-risk abnormal behaviors.

[0154] Based on the hierarchical abnormal behavior task list and the preset virtual safety guard area, the system automatically selects the most suitable warning algorithm for the current scenario, optimizes and configures the parameters according to the scenario characteristics, and generates a differentiated warning plan. Subsequently, parallel analysis and processing are carried out, and a multi-model fusion method is used to cross-verify the accuracy of abnormal judgment, generating an initial warning data set with credibility indicators. This data set contains key information such as abnormal type, location, time, and severity, providing a basis for subsequent warning information organization.

[0155] Based on the initial warning data set and the adaptive reminder rules, the system hierarchically organizes the warning information, adds notification marks according to the preset notification control logic, and constructs a warning information packet with a multi-level protection mechanism. For warning information at different levels, the system makes a fine-grained division of response permissions, assigns differentiated response permissions to different roles (such as family members, community caregivers, medical staff), and forms an initial response permission rule set.

[0156] Combined with the initial response permission rule set and the requirements of specific monitoring scenarios, the system establishes a response process state machine model, defines state transition rules, and forms a dynamic response state diagram. Based on this state diagram, the system evaluates the potential risks of each response operation, sets corresponding safety thresholds, and generates response risk control parameters. These parameters ensure that the system can maintain a safety margin when performing response operations, avoiding situations of over-intervention or insufficient response.

[0157] Finally, the system generates hierarchical warning signals according to the response risk control parameters, and pushes the warning instructions to relevant terminal devices in a distributed manner. The system also designs a standardized response process protocol and conducts effectiveness verification, and finally forms a complete hierarchical warning information and response permission strategy to provide comprehensive protection for the home safety of the elderly.

[0158] In a specific embodiment, the classifying the abnormal behaviors in different scenarios and generating warning signals according to the activity feature tags, the virtual safety guard area, and the adaptive reminder rules to obtain hierarchical warning information and response permission strategy may specifically include the following steps:

[0159] Perform scenario analysis and feature extraction on the behavior data to obtain a set of behavior feature vectors, and perform an abnormal behavior severity assessment and priority ranking according to the set of behavior feature vectors and the activity feature tags to obtain a hierarchical abnormal behavior task list;

[0160] Based on the hierarchical abnormal behavior task list and the virtual safety guard area, automatically select and configure warning algorithms to obtain a differentiated warning plan, and perform parallel analysis and accuracy verification of abnormal behaviors according to the differentiated warning plan to obtain an initial warning data set;

[0161] Perform warning information hierarchical organization and notification control marking based on the initial warning data set and the adaptive reminder rule to obtain a warning information packet with multi-level protection, and perform fine-grained division of response permissions and policy generation according to the warning information packet with multi-level protection to obtain an initial response permission rule set;

[0162] Perform response process state machine modeling and conversion rule definition based on the initial response permission rule set and monitoring scenario requirements to obtain a dynamic response state diagram, and perform response operation risk assessment and threshold setting according to the dynamic response state diagram to obtain response risk control parameters;

[0163] Generate and distribute push warning signals at different levels based on the response risk control parameters to obtain a distributed warning instruction set, and perform protocol design and effectiveness verification of the response process according to the distributed warning instruction set to obtain warning information and response permission policies at different levels.

[0164] Specifically, perform scenario analysis and feature extraction on behavior data, process the behavior data through a deep learning model, and extract feature vectors containing four dimensions: time, space, behavior, and environment. Among them, time features (such as behavior duration and occurrence time), space features (such as position coordinates and movement trajectories), behavior features (such as posture changes and action speeds), and environmental features (such as indoor temperature and lighting conditions, etc.). These features together constitute a behavior feature vector set, comprehensively describing the behavior state of the elderly. Perform abnormal behavior severity assessment according to the behavior feature vector set and activity feature labels, and calculate the severity score using the multi-index analytic hierarchy process. The indicators include the degree of deviation from the normal behavior pattern, duration, position risk coefficient, abnormal degree of physiological parameters, etc. Perform priority sorting based on the severity score to form a hierarchical abnormal behavior task list, including three levels: high risk, medium risk, and low risk.

[0165] Automatically select a warning algorithm based on the hierarchical abnormal behavior task list and the virtual safety guard area. The system selects the most suitable warning algorithm from the algorithm library according to the regional characteristics and risk levels of the abnormal behavior. The algorithm selection process can be expressed as: , where is the selected optimal algorithm, is the set of candidate algorithms, is the parameter of the virtual safety guard area, is the risk level, is the evaluation function, used to calculate the algorithm in the area for the risk level The applicability score. For example, for fall detection in high-risk areas such as bathrooms, the system may choose a deep learning-based pose analysis algorithm; while for abnormal behavior detection in open areas such as living rooms, it may choose an abnormal behavior recognition algorithm based on trajectory analysis. Optimize the parameter configuration for the selected algorithm using the Bayesian optimization method: , where is the optimal parameter set, is the candidate parameter set, is the expected value of parameter performance, calculated through cross-validation. Through parameter optimization and configuration, a differentiated early warning plan suitable for the current scenario is obtained. Based on the differentiated early warning plan, perform parallel analysis and processing of abnormal behaviors, using a multi-model parallel computing framework to simultaneously run different models to evaluate abnormal behaviors. The parallel processing can be expressed as:

[0166] , where represents the analysis result of the th model, that is, the evaluation score. Check the accuracy of the parallel processing results using a weighted voting mechanism: , where is the credibility coefficient of the model. When exceeds the preset threshold, the system confirms the existence of abnormal behavior and generates an initial early warning data set. Each element in the initial early warning data set contains key information such as abnormal type, location, time, severity, and credibility.

[0167] Based on the initial early warning data set and the adaptive reminder rules, organize the early warning information hierarchically, using a three-layer structure: the core layer contains key early warning information, the extended layer contains detailed context information, and the additional layer contains auxiliary information and suggested measures. Combine the adaptive reminder rules to perform notification control marking on the early warning information, adding recipient permission marking, timeliness marking, and privacy protection marking. The marking process can be expressed as: , where represents the th marking added to the information . Through the marking process, an early warning information package with a multi-level protection mechanism is obtained

[0168] . Based on the early warning information package, perform fine-grained division of response permissions, define a permission matrix for different roles. Each element in the permission matrix represents the permission level of the role for the early warning information, and the value set of the element is {0, 1, 2, 3} (0 means no permission, 1 - 3 means different levels of access and operation permissions). Generate policy rules according to the permission matrix and the early warning level. Each rule can be expressed as a conditional statement in the form of "if (condition) then (action)". These rules together constitute the initial response permission rule set.

[0169] Model the state machine of the response process based on the initial response permission rule set and the monitoring scenario requirements. Define the state set Q = {q0, q1,..., qn}, where q0 is the initial state and other states represent different stages in the response process. Define the state transition function δ(q, e) = q', which means transitioning to state q' after receiving event e in state q. The state transition rules can be represented as a set of conditions C = {c1, c2,..., cm}, where each condition cm defines the triggering condition for a specific state transition. Through the state set Q, the transition function δ, and the condition set C, construct the dynamic response state graph G = (Q, End, δ, q0, Fs), where End is the set of events and Fs is the set of termination states. Conduct a risk assessment of the response operation based on the dynamic response state graph, identify potential risk points using the fault tree analysis method, and calculate the operation risk score:

[0170] , where represents the operation, represents the operation the probability of a failure occurring, represents the severity of the impact of the failure. Set safety thresholds for high-risk operations:

[0171] , where is the minimum threshold, is the safety factor. These thresholds together constitute the response risk control parameter set.

[0172] Generate early warning signal grading based on the response risk control parameters, and construct a five-level early warning signal system: five levels of emergency, high warning, medium warning, low warning, and prompt information. The early warning level determination function is: , where represents the early warning level, is the early warning data, is the response risk control parameter set, It is an evaluation function, and the evaluation result is 0 or 1. The evaluation function includes multiple evaluation indicators, and the evaluation results of different evaluation indicators are summed. When the evaluation result of each evaluation indicator is greater than the corresponding parameter in the response risk control parameter set, the value of the evaluation function is 1. For warning signals at different levels, a distributed push strategy is adopted: for the emergency level, it is pushed to all relevant terminals simultaneously; for the high warning level, it is preferentially pushed to the terminals of the caregivers in the vicinity; for other levels, it is pushed directionally according to the roles and locations of the recipients. Through this distributed push mechanism, a distributed warning instruction set is formed. Based on the warning instruction set, a standardized response process protocol is designed, including three stages: confirmation of receipt, response action, and result feedback. The protocol is verified for effectiveness by using simulation tests and statistical analysis methods, and key performance indicators are calculated: average response time, success rate, false alarm rate, etc. Through comprehensive evaluation, a complete hierarchical warning information and response permission strategy is finally formed to provide comprehensive protection for the home safety monitoring of the elderly.

[0173] Based on the integration and analysis of the behavior trajectory analysis report, the response permission strategy, the processing records of the hierarchical warning information, and the external medical resource information, a dynamic intervention strategy is obtained;

[0174] Specifically, the time series correlation and pattern extraction are carried out on the behavior trajectory analysis report and the processing records of the hierarchical warning information to mine the internal correlation between the daily behaviors and health conditions of the elderly, and a feature matrix comprehensively reflecting the health risks of the elderly is obtained. This matrix contains multi-dimensional features such as time, space, behavior, and physiology, and can characterize the abnormal patterns of the health status of the elderly. Based on the health risk feature matrix and the response permission strategy, the correlation between different abnormal states is calculated by using the multi-factor correlation analysis method, and systematic classification is carried out to form an initial risk scenario set containing multiple risk scenarios.

[0175] Based on the initial risk scenario set, a model of the change trend graph of the health status of the elderly is constructed, and cluster analysis is carried out on different types of health status changes to form a health status change graph. This change graph clearly shows the evolution path and trend characteristics of the health status of the elderly, providing a scientific basis for intervention decision-making. Based on the health status change graph, a systematic medical needs assessment is carried out, and the urgency of different situations is accurately divided to form a preliminary intervention needs division result, laying a foundation for subsequent precise intervention.

[0176] Based on the preliminary intervention needs division result, a deep learning method is used to identify high-risk states, and the intervention effect is analyzed by combining historical intervention records. An intervention needs correlation matrix is constructed to comprehensively represent the correlation and priority between different intervention needs. Based on the intervention needs correlation matrix, an emergency response mode model is constructed, and multi-scenario scenario deduction is carried out to form an intervention flow chart atlas, providing a clear intervention decision-making path and execution process.

[0177] Based on the intervention flow chart and external medical resource information, set the medical resource allocation threshold, plan the optimal rescue path, and form a set of resource allocation strategies. This set of strategies includes resource allocation plans under different emergency levels to ensure the efficient utilization of medical resources. Based on the set of resource allocation strategies, construct a dynamic intervention environment, optimize the rescue team configuration, and finally form a comprehensive, accurate, and efficient dynamic intervention strategy to provide timely and appropriate safety guarantees for the elderly.

[0178] In a specific embodiment, the integration and analysis of the behavior trajectory analysis report, the response permission policy, the processing record of the hierarchical warning information, and the external medical resource information to obtain the dynamic intervention strategy may specifically include the following steps:

[0179] Perform temporal correlation and pattern extraction on the behavior trajectory analysis report and the processing record of the hierarchical warning information to obtain a health risk feature matrix, and perform abnormal state correlation calculation and classification based on the health risk feature matrix and the response permission policy to obtain an initial risk scenario set;

[0180] Based on the initial risk scenario set, conduct trend graph modeling and type clustering analysis on the changing trend of the elderly's health status to obtain a health status change graph, and conduct medical needs assessment and emergency level division based on the health status change graph to obtain a preliminary intervention needs division result;

[0181] Based on the preliminary intervention needs division result, conduct high-risk state identification and intervention effect analysis to obtain an intervention needs correlation matrix, and conduct an emergency response mode modeling and plan deduction based on the intervention needs correlation matrix to obtain an intervention flow chart;

[0182] Based on the intervention flow chart and external medical resource information, set the medical resource allocation threshold and plan the rescue path to obtain a set of resource allocation strategies, and construct a dynamic intervention environment and configure the rescue team based on the set of resource allocation strategies to obtain a dynamic intervention strategy.

[0183] Specifically, perform temporal correlation and pattern extraction on the behavioral trajectory analysis report and the processing records of the classification warning information, and use time series analysis methods to conduct correlation analysis on the behavioral trajectory analysis report and the warning processing records. The system first serializes the behavioral trajectory analysis report according to the timestamp to form a temporal behavior sequence, and each element in the temporal behavior sequence represents a behavioral feature vector, including information such as location, action type, and duration. At the same time, form a temporal warning sequence from the processing records of the classification warning information, and each element in the temporal warning sequence represents a warning event feature vector, including information such as warning type, level, and processing method. Calculate the temporal correlation degree of the two sequences through the dynamic time warping algorithm. This algorithm calculates their similarity by finding the optimal alignment path between two time series, and can accurately capture their correlation even if the sequence lengths are different or there are time delays.

[0184] Based on the results of the temporal correlation analysis, use the non-negative matrix factorization method to extract the health risk patterns and construct a health risk feature matrix. Assume that the comprehensive feature data matrix is , with a dimension of , where represents the number of samples, represents the feature dimension. The non-negative matrix factorization method decomposes Xty into the product of two non-negative matrices: , where is a -dimensional basis matrix, is a -dimensional coefficient matrix, is the number of patterns extracted, which is usually much smaller than and . Each column of the basis matrix represents a health risk pattern, and the coefficient matrix represents the weight distribution of each pattern. Through this decomposition, clear health risk patterns can be extracted from complex behavioral and warning data to form a health risk feature matrix.

[0185] According to the health risk feature matrix and the response permission policy, perform abnormal state correlation calculation and classification, and use the association rule mining algorithm to calculate the correlation between abnormal states. Identify strong association rules by calculating support and confidence:

[0186] , represents the probability of the simultaneous occurrence of abnormal states and ;

[0187] , where

[0188] represents that in the abnormal state Under the condition of occurrence, The conditional probability of occurrence. For association rules that meet the threshold conditions: Greater than or equal to a preset association threshold and Greater than or equal to a preset confidence threshold, incorporate them into the abnormal state association network.

[0189] Combined with the response permission policy matrix, the elements in the permission policy matrix Represent the response permission level for the abnormal state Of Applicability. By calculating the weighted association degree:

[0190] , further optimize the abnormal state association network. Use the spectral clustering algorithm to classify the abnormal states to obtain the abnormal state classification results. Finally, based on the classification results and association strength, construct an initial risk scenario set, where each risk scenario contains a group of highly associated abnormal states and corresponding description information.

[0191] Based on the initial risk scenario set, conduct trend graph modeling and type clustering analysis on the changing trend of the elderly's health status, and use a time series graph neural network (TGNN) to construct a health status trend graph model. Consider each scenario in the risk scenario set as a node in the graph, construct directed edges through the transition probability within the time span to form an initial trend graph G0=(V,E,A), where V is the set of nodes, E is the set of edges, A is the adjacency matrix, and a_ij represents the probability of transferring from scenario s_i to scenario s_j.

[0192] Use a graph convolutional network (GCN) for feature extraction. The function formula of the graph convolutional network is:

[0193] , where Is the node feature matrix of the Layer, Is the degree matrix, Is the learnable weight matrix, Is the activation function. By stacking multiple layers of GCN, extract the high-order feature representations of the nodes to capture the complex change patterns of the health status. Combine the time convolutional network (TCN) to capture the time series features of the output features of the last layer of GCN. TCN captures the change features at different time scales through one-dimensional convolution and dilated convolution. Finally, through the fusion of a multi-layer perceptron, obtain the fusion features.

[0194] Based on the fusion features, the spectral clustering algorithm is used to cluster the types of health status change patterns. A similarity matrix is constructed, and the k-medoids clustering algorithm is applied to divide the health status change patterns into multiple categories. Through visualization technology, the clustering results are mapped to a two-dimensional plane to construct a health status change graph, which intuitively shows the evolution path and transition relationship between different health states.

[0195] Based on the health status change graph, medical needs assessment and urgency classification are carried out, and a multi-factor scoring model is used for medical needs assessment. An evaluation index system is designed, including multiple dimensions such as physiological conditions, degree of behavioral abnormalities, and warning frequencies. For each state node in the health status change graph, a weight formula is used to calculate the comprehensive urgency, that is, the scores of each state on the corresponding indicators are weighted and summed. Based on the comprehensive urgency, a threshold is divided to divide the health status into multiple urgency levels, such as "normal", "need attention", "need intervention", "urgent intervention", etc. Combining the urgency level and the health status change graph, a preliminary intervention needs division result is formed, providing a scientific basis for the formulation of subsequent intervention strategies.

[0196] Based on the preliminary intervention needs division result, high-risk state identification and intervention effect analysis are carried out, and the deep reinforcement learning method is used for high-risk state identification. The health status space is modeled as a Markov decision process (MDP), the state space is the set of nodes in the health status change graph, and the action space is the set of possible intervention methods. A reward function based on health risk is designed:

[0197] , where represents the health status before intervention, represents the health status after intervention, represents the state 's urgency, represents the intervention action 's cost, and are trade-off parameters. The optimal intervention strategy is learned through a deep Q-network, and the high-risk state set is identified through Q-value analysis. Combining historical intervention records, the intervention effect is analyzed, and based on the preset weight, a weight formula is used to calculate the intervention effect score. An intervention needs association matrix is constructed, and the matrix elements represent the association strength between different intervention needs, which is calculated by the cosine similarity of eigenvectors.

[0198] Based on the intervention requirement correlation matrix, conduct emergency response mode modeling and scenario deduction. Use a graph neural network to model the emergency response mode. Consider the intervention requirement correlation matrix as the adjacency matrix of a weighted graph, and construct an intervention requirement graph, where nodes represent intervention requirements and edges represent the correlation relationships between requirements. Apply a graph attention network to extract node features, and use the features extracted by the graph attention network to construct a decision tree model for response scenario deduction. For each intervention requirement node, based on its feature vector, derive a series of possible intervention scenarios. Each scenario includes elements such as the intervention timing, intervention means, and implementing roles. Through the Monte Carlo tree search algorithm, simulate the execution processes of multiple intervention scenarios on the decision tree, and evaluate the expected effects and risks of each scenario. Through multiple rounds of deduction and evaluation, select the optimal set of intervention scenarios, and construct an intervention flow chart spectrum, where nodes represent intervention decision points or actions, and edges represent the decision-making process. The intervention flow chart spectrum adopts a directed acyclic graph structure, clearly showing the complete intervention process from anomaly detection to emergency response, providing executable guidance for actual operations.

[0199] Based on the intervention flow chart spectrum and external medical resource information, set the medical resource allocation threshold and plan the rescue path. Set the resource allocation threshold through a multi-objective optimization method. Classify medical resources into multiple categories, including medical staff, first aid equipment, drugs, etc. For each node in the intervention flow chart spectrum, set a resource requirement vector according to its intervention requirements. Combine the external medical resource availability information to construct resource allocation constraint conditions: the demand is less than or equal to the external medical resource availability information. Design a multi-objective optimization problem, considering both resource utilization efficiency and intervention effects, and determine the medical resource allocation threshold vector through Pareto optimal solution analysis. When the demand exceeds the threshold, trigger the corresponding level of resource allocation.

[0200] Use a heuristic algorithm to plan the rescue path, modeled as a multi-constraint shortest path problem. Consider the set of geographical locations, including the residences of the elderly, medical institutions, the locations of rescue personnel, etc. Construct a weighted graph, where the edge weights include multi-dimensional information such as distance, time, and traffic conditions. For an emergency, it is necessary to find the optimal path from the rescue source point to the destination, minimizing the weighted cost, including the time, distance, and other cost factors of the path. Taking time and distance as examples, use a heuristic search algorithm to solve the shortest path problem, and the heuristic function is designed as: , where represents the heuristic estimate value, is the heuristic estimate based on time and is the heuristic estimate based on distance, and are balance parameters. Based on different emergency levels and resource types, generate multiple sets of rescue path plans to form a resource allocation strategy set. Each strategy includes information such as the resource allocation threshold, allocation order, and rescue path.

[0201] Based on the resource allocation strategy set, a dynamic intervention environment is constructed and rescue teams are configured. The digital twin technology is used to construct the dynamic intervention environment. Information such as the living environment of the elderly, the distribution of medical resources, and the transportation network in the physical space is mapped into the virtual space to form a digital twin model, including the environmental topological structure, the elderly and related objects, resource entities, and status monitoring points. The digital twin model is updated through real-time data synchronization to ensure that the virtual environment is consistent with the actual situation. Based on the resource allocation strategy set and the digital twin environment, a combinatorial optimization method is used to configure the rescue teams. Let there be a set of rescue personnel, and each person has a capability vector representing their professional capabilities in different intervention tasks. The team configuration problem is modeled as a bipartite graph matching problem, and the optimization goal is: to maximize the matching degree between the allocated personnel and the capabilities and requirements of the allocated tasks. The optimal personnel allocation plan is solved through the Hungarian algorithm, and combined with the resource allocation strategy, a complete dynamic intervention strategy is formed, including the intervention timing strategy, the resource allocation strategy, the personnel configuration strategy, and the intervention process strategy. The dynamic intervention strategy evaluates the current state in real time through a rule engine and adaptively adjusts the intervention parameters to ensure the accuracy and timeliness of the intervention measures, providing comprehensive and intelligent home safety guardianship for the elderly.

[0202] Update the personalized safety risk assessment model according to the dynamic intervention strategy to obtain the target safety guardianship model;

[0203] Specifically, a systematic correlation analysis and in-depth pattern mining are carried out on the dynamic intervention strategy and historical intervention records to construct a sequence diagram reflecting the time-series effect of the intervention measures, providing a data basis for the evaluation of the intervention measures. Combining the data on the changes in the health status of the elderly, the effectiveness and adaptability of different intervention measures are evaluated, and a quantitative intervention effect measurement table is established to provide a scientific basis for subsequent optimization of the intervention strategy. Through the correlation analysis between the intervention effect measurement table and the behavioral characteristics of the elderly, a complex correlation network between risk factors is established to reveal the causal relationship and influence path between risk factors. Based on the correlation network, health risk reasoning and probability calculation are carried out to construct a health risk probability tree to realize the prediction and quantitative evaluation of potential health risks.

[0204] Based on the health risk probability tree, the optimal allocation of guardianship resources is carried out. Considering the risk severity, resource availability, and intervention effect comprehensively, a preliminary guardianship strategy plan is formulated. A multi-dimensional feasibility evaluation and priority ranking are carried out on the preliminary plan to form a hierarchical guardianship strategy set with a hierarchical structure, which is suitable for different risk levels and resource conditions. Based on the hierarchical guardianship strategy set, specific guardianship rules are dynamically generated, and the consistency between the rules is ensured through consistency detection to form a complete target guardianship rule library. According to the rule library, the automated choreography of the daily guardianship process and the construction of decision-making logic are carried out to generate an executable intelligent guardianship plan, realizing the intelligence and automation of the guardianship process.

[0205] Dynamically adjust the risk assessment parameters based on the intelligent guardianship solution, update the monitoring conditions according to the real-time monitoring data, and form a temporary safety assessment strategy. Apply the temporary strategy to the global optimization process of the personalized safety risk assessment model. By balancing the weights of various factors and adjusting the parameter relationships, finally obtain a target safety guardianship model with high precision and high adaptability, providing comprehensive protection for the home safety of the elderly.

[0206] In a specific embodiment, updating the personalized safety risk assessment model according to the dynamic intervention strategy to obtain a target safety guardianship model may specifically include the following steps:

[0207] Associate and perform pattern mining on the dynamic intervention strategy and historical intervention records to obtain an intervention effect sequence diagram, and perform effectiveness evaluation and adaptability detection of the intervention measures based on the intervention effect sequence diagram and the changes in the health status of the elderly to obtain an intervention effect measurement table;

[0208] Perform risk factor association analysis based on the intervention effect measurement table and the behavioral characteristics of the elderly to obtain a risk factor association network, and perform potential health risk reasoning and probability estimation based on the risk factor association network to obtain a health risk probability tree;

[0209] Perform optimized allocation of guardianship resources and monitoring planning based on the health risk probability tree to obtain a preliminary guardianship strategy plan, and perform feasibility evaluation and priority ranking of the guardianship measures based on the preliminary guardianship strategy plan to obtain a hierarchical guardianship strategy set;

[0210] Perform dynamic generation and consistency detection of guardianship rules based on the hierarchical guardianship strategy set to obtain a target guardianship rule library, and perform automated choreography of the daily guardianship process and construction of a decision tree based on the target guardianship rule library to obtain an intelligent guardianship solution;

[0211] Perform dynamic adjustment of the risk assessment parameters and update of the monitoring conditions based on the intelligent guardianship solution to obtain a temporary safety assessment strategy, and perform global optimization and balanced adjustment of the personalized safety risk assessment model according to the temporary safety assessment strategy to obtain a target safety guardianship model.

[0212] Specifically, associate and perform pattern mining on the dynamic intervention strategy and historical intervention records, and use time series data mining methods to analyze the execution effect of the intervention strategy. First, represent the dynamic intervention strategy as a sequence, where each intervention strategy contains information such as the intervention type, execution time, and execution entity. At the same time, represent the historical intervention records as a time series, where each record contains information such as the intervention execution situation, the elderly's reaction, and changes in health indicators. Use the time series similarity calculation method to analyze the matching degree between the intervention strategy and the execution record: , where denotes the temporal distance calculation function, denotes the intervention strategy, denotes the execution record, which can be calculated using the dynamic time warping algorithm, is a scale parameter that controls the similarity decay rate. Using the sequential pattern mining algorithm, frequently occurring intervention patterns are identified from the intervention records. An intervention pattern represents a sequence of intervention behaviors, and the arrow represents the temporal relationship. For each pattern, the frequency of the pattern's occurrence in the historical records is used as the support. Patterns with a support greater than a preset threshold are selected as representative intervention patterns. Based on the mined intervention patterns, an intervention effect sequence diagram is constructed. The nodes in the diagram represent intervention behaviors and health states, the edges represent state transition relationships, and the edge weights represent transition probabilities, which is the number of times of transitioning from state A1 to state B1 divided by the total number of times state A1 appears. The intervention effect sequence diagram clearly shows the sequential relationships between different intervention behaviors and the impact path on the health state through a directed graph structure.

[0213] According to the intervention effect sequence diagram and the changes in the health state of the elderly, the effectiveness evaluation and adaptability detection of the intervention measures are carried out using a multi-index evaluation method. Define the health state change vector

[0214] , where and represent the health state vectors before and after the intervention, respectively. The health state vector includes multi-dimensional features such as physiological indicators, activity ability, and cognitive state. For the intervention measures, calculate the effectiveness score of the intervention measures: multiply the weight vector by the health state change vector. The adaptability detection uses the cross-validation method to divide the elderly into different types of groups based on characteristics such as age, health status, and living habits. For each group and intervention measure, calculate the adaptability score: divide the mean of the effectiveness scores in the group by the standard deviation. The higher this adaptability score, the better the adaptability of the intervention measure in this group. Combining the evaluation results of effectiveness and adaptability, an intervention effect measurement table is constructed. The measurement table is in matrix form, with rows representing intervention measures and columns representing evaluation indicators. The matrix elements represent the scores of the intervention on the indicators. Through normalization processing, the scores of each indicator are mapped to the [0,1] interval, and finally a standardized intervention effect measurement table is obtained, providing a quantitative basis for subsequent analysis.

[0215] Based on the intervention effect measurement table and the elderly behavior characteristic data, a risk factor association analysis is carried out. The association analysis identifies the association relationships between risk factors by calculating the mutual information value and conditional dependence probability between risk factors. Let the association strength between risk factors and be , and the calculation method is: ; where represents the mutual information function, represents the information entropy function. This formula calculates the normalized mutual information between two risk factors, reflecting the degree of association between them. The higher the mutual information value, the stronger the association between the two risk factors. Based on the association analysis results, a risk factor association network is constructed. The association network is a weighted undirected graph, where nodes represent risk factors, edges represent the association relationships between risk factors, and the weights of the edges represent the association strength. Based on the risk factor association network, potential health risk reasoning is carried out. The health risk reasoning adopts a Bayesian network model, and calculates the health risk probability under different combinations of risk factors through conditional probability. Let the health risk event be , and the set of risk factors be , then the health risk probability is calculated as: ; this formula represents the probability of the occurrence of the health risk event under the condition of knowing the states of each risk factor. The conditional probability is obtained through the integration of historical data learning and expert knowledge, reflecting the degree of influence of risk factors on health risks. By enumerating different state combinations of risk factors, the corresponding health risk probabilities are calculated to construct a health risk probability tree. The health risk probability tree is a tree-like structure, where the root node represents the initial state, the leaf nodes represent the final health risk states, the intermediate nodes represent the combinations of risk factor states, the branches represent the state transition paths, and the cumulative product of the probabilities on each path represents the occurrence probability of that path.

[0216] Based on the health risk probability tree, the optimal allocation of monitoring resources is carried out. Monitoring resources include sensing devices, caregivers, medical resources, etc. The resource allocation is achieved by solving a multi-objective optimization problem to balance the risk control effect and resource cost. Let the monitoring resource allocation plan be , and its objective function is:

[0217] where is the balance parameter, represents the risk control effect function of the plan , and represents the resource cost function of the plan . The goal is to minimize , that is, to maximize the risk control effect while controlling the cost. Based on the results of the optimal resource allocation, a monitoring plan is formulated to clarify the deployment locations, monitoring frequencies, and data collection strategies of various monitoring devices, forming a preliminary monitoring strategy plan. The feasibility of the preliminary monitoring strategy plan is evaluated, considering the acceptance of the elderly, the difficulty of technical implementation, and environmental adaptability. Let the weight formula be used as the evaluation function to calculate the feasibility evaluation results. According to the feasibility evaluation results, the monitoring measures are prioritized to form a hierarchical monitoring strategy set. The prioritization considers the risk control effect, urgency, and resource requirements of the measures to ensure that key monitoring measures are implemented first.

[0218] Based on the hierarchical monitoring strategy set, dynamic generation of monitoring rules is carried out. The monitoring rules include three parts: triggering conditions, execution operations, and feedback mechanisms. Rule generation adopts template filling and rule induction methods to transform the strategies into executable rule statements. Consistency detection is performed on the generated rules to identify conflicts and overlaps between the rules and optimize and adjust them. Consistency detection uses rule comparison and logical reasoning to test the behavioral consistency of the rule set in various situations to ensure that contradictory decisions will not be generated. Based on the optimized rule set, a target monitoring rule library is formed. Based on the target monitoring rule library, automated choreography of the daily monitoring process is carried out. Process choreography constructs a complete monitoring workflow by defining the execution order, conditional branches, and information flow of monitoring activities. The goal of process choreography is to optimize the execution efficiency of activities and the risk response timeliness. Through process modeling and optimization, a decision tree is constructed to clarify the decision-making paths and execution strategies in various situations, forming an intelligent monitoring solution. Based on the intelligent monitoring solution, dynamic adjustment of risk assessment parameters is carried out. Parameter adjustment updates the key parameters in the risk assessment model by analyzing monitoring data and risk event feedback. Let the parameter set of the risk assessment model be , and the parameter update function is: ; where is the learning rate, represents the loss gradient of parameter . This formula realizes the gradient descent update of the parameters, enabling the model to better adapt to new data and situations. At the same time, the monitoring conditions are updated, the threshold settings and triggering rules are adjusted, and a temporary safety assessment strategy is formed. According to the temporary safety assessment strategy, global optimization of the personalized safety risk assessment model is carried out. Global optimization improves the accuracy and adaptability of the model through model structure adjustment and parameter retraining. The cross-validation method is adopted in the optimization process to ensure the stability of the model in different elderly populations and different scenarios. Equalization adjustment is performed on the optimized model to balance sensitivity and specificity, ensuring that the model can identify risks in a timely manner without generating too many false alarms, and finally forming a target safety monitoring model.

[0219] This embodiment realizes precise monitoring by constructing a risk assessment model, avoiding the one-size-fits-all problem of traditional monitoring systems. Through the environmental perception network constructed by multi-type sensors, intelligent identification of abnormal behaviors and risk grading early warnings are realized, greatly improving the accuracy and timeliness of early warnings. According to the analysis of behavioral trajectories, safety areas are automatically divided and intelligently configured to form virtual safety guardianship areas, realizing key monitoring of high-risk areas. Self-optimize according to the intervention effect and changes in health status, continuously update the risk assessment model, and improve the long-term monitoring effect. Combine external medical resource information to formulate dynamic intervention strategies to achieve efficient allocation of medical resources in case of emergencies. Adopt a hierarchical response permission strategy to protect the privacy and dignity of the elderly to the greatest extent while ensuring safety.

[0220] Example 2;

[0221] Please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, refer to the description in Embodiment 1. The method for home safety monitoring of retired elderly people based on the Internet of Things includes:

[0222] S1. Deploy multiple types of sensors in the elderly people's home environment for all-weather data collection and analysis to obtain behavior data with activity feature tags;

[0223] S2. Conduct correlation analysis based on the behavior data and the elderly people's health record information to obtain a personalized safety risk assessment model;

[0224] S3. According to the personalized safety risk assessment model, conduct real-time monitoring and pattern recognition of the elderly people's daily activities to obtain a behavior trajectory analysis report;

[0225] S4. Based on the behavior trajectory analysis report, conduct safety level division and intelligent configuration of the elderly people's living areas to obtain a virtual safety guardianship area and an adaptive reminder rule;

[0226] S5. According to the activity feature tags, the virtual safety guardianship area, and the adaptive reminder rule, conduct hierarchical classification processing of the behavior data and generate early warning signals to obtain hierarchical early warning information and a response permission policy;

[0227] S6. Integrate and analyze the behavior trajectory analysis report, the response permission policy, the processing records of the hierarchical early warning information, and external medical resource information to obtain a dynamic intervention strategy, and update the personalized safety risk assessment model according to the dynamic intervention strategy to obtain a target safety monitoring model.

[0228] This embodiment also provides a home safety monitoring device for retired elderly people based on the Internet of Things. The home safety monitoring device for retired elderly people based on the Internet of Things includes a memory and a processor. When computer-readable instructions stored in the memory are executed by the processor, the processor executes the steps of the home safety monitoring method for retired elderly people based on the Internet of Things in the above embodiments.

[0229] This embodiment also provides a computer-readable storage medium. The computer-readable storage medium can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer executes the steps of the home safety monitoring method for retired elderly people based on the Internet of Things.

[0230] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula that is closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0231] The above is only the preferred embodiment of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions within the idea of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A home safety monitoring system for retired elderly people based on the Internet of Things, characterized in that: include: Collection and processing module: deploy multiple types of sensors in the elderly’s home environment to collect and analyze data around the clock, and obtain behavioral data with activity feature labels; Model building module: performing correlation analysis based on the behavioral data and the health record information of the elderly to obtain a personalized safety risk assessment model; Analysis module: Based on the personalized safety risk assessment model, the daily activities of the elderly are monitored in real time and pattern recognized to obtain a behavior trajectory analysis report; Division module: Based on the behavior trajectory analysis report, the living area of ​​the elderly is divided into security levels and intelligently configured to obtain virtual security guard areas and adaptive reminder rules; Classification module: classifies the behavior data and generates warning signals according to the activity feature labels, the virtual security guard area and the adaptive reminder rules, and obtains graded warning information and response authority strategies; Optimization module: based on the behavior trajectory analysis report, the response authority strategy, the processing record of the graded warning information and the external medical resource information, an integrated analysis is performed to obtain a dynamic intervention strategy, and the personalized safety risk assessment model is updated according to the dynamic intervention strategy to obtain a target safety monitoring model; The personalized safety risk assessment model is obtained by performing correlation analysis based on the behavior data and the health record information of the elderly, including: Performing structured analysis and feature extraction on the health record information of the elderly to obtain a health status feature vector, and identifying health risk factors based on the health status feature vector to obtain an initial health risk rating; Based on the initial health risk rating and the behavior data, a correlation analysis is performed on the relationship between the health status and the behavior pattern to obtain a risk association rule set and a health behavior characteristic model, and according to the risk association rule set and the health behavior characteristic model, the health status and the behavior pattern are cross-validated and weighted to obtain a health behavior risk association map; Calculating the importance and prioritizing the risk factors based on the health behavior risk association map to obtain a risk factor weight matrix, and quantifying the levels of potential risk scenarios based on the risk factor weight matrix and the activity feature labels to obtain an initial risk assessment matrix; Based on the initial risk assessment matrix, a time series pattern analysis is performed on the risk prediction to obtain a set of risk evolution rules, and according to the set of risk evolution rules, a Markov chain modeling and risk transfer probability calculation are performed on the initial risk assessment matrix to obtain a dynamic risk probability distribution; Based on the dynamic risk probability distribution, risk prediction accuracy assessment and threshold optimization are performed to obtain a risk model adjustment parameter set, and based on the risk model adjustment parameter set, risk assessment multi-feature fusion and model training verification are performed to obtain a personalized safety risk assessment model.

2. The home safety monitoring system for retired elderly people based on the Internet of Things according to claim 1 is characterized in that: The above method deploys multiple types of sensors in the elderly's home environment to collect and analyze data around the clock, and obtains behavioral data with activity feature labels, including: Conduct spatial planning and sensor demand analysis on the elderly’s home environment, obtain a sensor distribution plan, and deploy multiple types of sensor devices based on the sensor distribution plan to build an environmental perception network; Continuously collecting environmental parameters and activity signals through the environmental perception network to obtain an original perception data stream, and performing signal filtering and timing alignment processing on the original perception data stream to obtain cleaned structured data; Based on the structured data, spatiotemporal feature calculation and behavior sequence segmentation are performed to obtain a basic behavior unit set, and a pattern recognition algorithm and context association analysis are applied to the basic behavior unit set to obtain behavior semantic features; Activity classification and attribute labeling are performed according to the behavior semantic features to obtain preliminary behavior labels, and the preliminary behavior labels are verified and enhanced in combination with the historical behavior pattern library to obtain behavior data with activity feature labels.

3. The home safety monitoring system for retired elderly people based on the Internet of Things according to claim 1 is characterized in that: According to the personalized safety risk assessment model, real-time monitoring and pattern recognition are performed on the daily activities of the elderly to obtain a behavior trajectory analysis report, including: Real-time collection and signal processing of elderly activity data are performed to obtain an original behavior data stream, and normal behavior definition and abnormality detection are performed on the elderly activities based on the original behavior data stream and the personalized safety risk assessment model to obtain an activity status marking result; Based on the activity state marking results, time and space feature extraction and activity type recognition are performed on the behavior events to obtain a multi-dimensional behavior feature set, and based on the multi-dimensional behavior feature set, the activity patterns of the elderly are analyzed and conventional behavior modeling is performed to obtain a behavior pattern library; Based on the behavior pattern library and historical activity data, a behavior time series model is constructed and trend analysis is performed to obtain a behavior change trajectory diagram, and behavior deviation calculation and risk scoring are performed based on the behavior change trajectory diagram and preset health and safety indicators to obtain a behavior risk measurement scale; Based on the behavior risk measurement table and the personalized safety risk assessment model, risk level classification and urgency ranking are performed to obtain a risk event priority queue, and monitoring tasks are intelligently allocated and monitored in parallel through the risk event priority queue to obtain a distributed monitoring task network; Based on the distributed monitoring task network, behavior data statistics aggregation and pattern extraction are performed to obtain structured behavior records, and activity association analysis and safety trend prediction are performed based on the structured behavior records to obtain a behavior trajectory analysis report.

4. The home safety monitoring system for retired elderly people based on the Internet of Things according to claim 1 is characterized in that: Based on the behavior trajectory analysis report, the living area of ​​the elderly is divided into security levels and intelligently configured to obtain a virtual security guard area and adaptive reminder rules, including: Performing spatiotemporal distribution statistics and frequency analysis on the behavior trajectory analysis report to obtain a behavior pattern feature matrix, and performing living area correlation calculation and classification based on the behavior pattern feature matrix and environmental factor data to obtain an initial area function set; Based on the initial regional function set, heat map modeling and regional clustering analysis are performed on the correlation between the activity trajectories of the elderly to obtain a behavior density distribution map, and safety risk assessment and level division are performed based on the behavior density distribution map to obtain a preliminary safety area division result; Based on the preliminary safety area division results, high-risk area identification and abnormal behavior analysis are performed to obtain a risk area association matrix, and daily activity pattern modeling and scenario deduction are performed based on the risk area association matrix to obtain a safety monitoring process map; Based on the safety monitoring process map and the health status indicators of the elderly, regional safety warning thresholds are set and monitoring points are deployed to obtain a regional monitoring strategy set, and based on the regional monitoring strategy set, a virtual safety guarding environment is constructed and sensors are configured to obtain a virtual safety guarding area; Based on the virtual safety guard area and the caregiver authority setting, the safety warning behavior is extracted and parameterized to obtain adaptive reminder rules.

5. The home safety monitoring system for retired elderly people based on the Internet of Things according to claim 1 is characterized in that: The step of performing graded classification processing and early warning signal generation on the behavior data according to the activity feature label, the virtual security guard area and the adaptive reminder rule to obtain graded early warning information and response authority strategy includes: Performing scenario analysis and feature extraction on the behavior data to obtain a behavior feature vector set, and performing abnormal behavior severity assessment and priority sorting according to the behavior feature vector set and the activity feature labels to obtain a graded abnormal behavior task list; Automatically selecting and configuring the early warning algorithm and parameters based on the hierarchical abnormal behavior task list and the virtual security guard area to obtain a differentiated early warning scheme, and performing parallel analysis and processing of abnormal behaviors and accuracy verification according to the differentiated early warning scheme to obtain an initial early warning data set; Based on the initial warning data set and the adaptive reminder rules, the warning information is hierarchically organized and notified to control the marking, so as to obtain a warning information package with multi-level protection, and the response authority is finely divided and the strategy is generated according to the multi-level protection warning information package to obtain an initial response authority rule set; Based on the initial response authority rule set and monitoring scenario requirements, response process state machine modeling and conversion rule definition are performed to obtain a dynamic response state diagram, and response operation risk assessment and threshold setting are performed according to the dynamic response state diagram to obtain response risk control parameters; Based on the response risk control parameters, early warning signals are generated and distributedly pushed to obtain a distributed early warning instruction set, and protocol design and performance verification of the response process are performed according to the distributed early warning instruction set to obtain hierarchical early warning information and response authority strategy.

6. The home safety monitoring system for retired elderly people based on the Internet of Things according to claim 1 is characterized in that: The dynamic intervention strategy is obtained by integrating and analyzing the behavior trajectory analysis report, the response authority strategy, the processing record of the graded warning information and the external medical resource information, including: Performing time series correlation and pattern extraction on the behavior trajectory analysis report and the processing record of the graded warning information to obtain a health risk feature matrix, and performing abnormal state correlation calculation and classification according to the health risk feature matrix and the response authority strategy to obtain an initial risk scenario set; Based on the initial risk scenario set, trend chart modeling and type cluster analysis are performed on the health status change trend of the elderly to obtain a health status change chart, and medical needs assessment and urgency classification are performed based on the health status change chart to obtain a preliminary intervention needs classification result; Based on the preliminary intervention demand classification results, high-risk state identification and intervention effect analysis are performed to obtain an intervention demand association matrix, and emergency response mode modeling and plan deduction are performed according to the intervention demand association matrix to obtain an intervention process map; Based on the intervention process map and external medical resource information, medical resource allocation thresholds are set and rescue path planning is performed to obtain a resource allocation strategy set, and based on the resource allocation strategy set, a dynamic intervention environment is constructed and a rescue team is configured to obtain a dynamic intervention strategy.

7. The home safety monitoring system for retired elderly people based on the Internet of Things according to claim 1 is characterized in that: The step of updating the personalized safety risk assessment model according to the dynamic intervention strategy to obtain a target safety monitoring model includes: The dynamic intervention strategy and historical intervention records are associated and pattern mined to obtain an intervention effect sequence diagram, and the effectiveness and adaptability of the intervention measures are evaluated and tested based on the intervention effect sequence diagram and changes in the health status of the elderly to obtain an intervention effect measurement table; Perform risk factor association analysis based on the intervention effect measurement table and the behavioral characteristics of the elderly to obtain a risk factor association network, and perform potential health risk reasoning and probability estimation based on the risk factor association network to obtain a health risk probability tree; Based on the health risk probability tree, optimal allocation of monitoring resources and monitoring planning are performed to obtain a preliminary monitoring strategy plan, and feasibility assessment and priority sorting of monitoring measures are performed according to the preliminary monitoring strategy plan to obtain a hierarchical monitoring strategy set; Dynamically generate and check the consistency of monitoring rules based on the hierarchical monitoring strategy set to obtain a target monitoring rule base, and automatically arrange the daily monitoring process and build a decision tree based on the target monitoring rule base to obtain an intelligent monitoring solution; Based on the intelligent monitoring solution, risk assessment parameters are dynamically adjusted and monitoring conditions are updated to obtain a temporary safety assessment strategy. According to the temporary safety assessment strategy, the personalized safety risk assessment model is globally optimized and balanced to obtain a target safety monitoring model.

8. Home safety monitoring equipment for retired elderly people based on the Internet of Things, characterized by: The home safety monitoring device for retired elderly people based on the Internet of Things comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory so that the IoT-based home safety monitoring device for retired elderly people implements the IoT-based home safety monitoring system for retired elderly people described in any one of claims 1-7.

9. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the home safety monitoring system for retired elderly people based on the Internet of Things described in any one of claims 1-7 is implemented.

Images