CN118039165B

CN118039165B - Cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system

Info

Publication number: CN118039165B
Application number: CN202410430626.XA
Authority: CN
Inventors: 廖陆枭; 吕从改
Original assignee: Liaoning Xinhao Medical Technology Co ltd
Current assignee: Liaoning Xinhao Medical Technology Co ltd
Priority date: 2024-04-11
Filing date: 2024-04-11
Publication date: 2024-06-11
Anticipated expiration: 2044-04-11
Also published as: CN118039165A

Abstract

The invention relates to the technical field of cardiovascular and cerebrovascular monitoring, in particular to an intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation. In the invention, the quality of physiological signal data is improved through Fourier transformation, the accuracy of detecting health trends and abnormal points is enhanced by combining a time sequence analysis and a statistical method, a behavior mode is identified by applying association rule mining and cluster analysis, health influence factors are mastered, a Monte Carlo simulation and decision tree analysis method is used for optimizing a rehabilitation strategy, the rehabilitation effect is improved, real-time health monitoring is realized by edge calculation and flow data analysis, risks are identified in time, and potential risks are accurately predicted by a graph theory and network topology method in heart brain network analysis, prediction modeling and prediction of a rehabilitation auxiliary process are carried out by the time sequence analysis, and a reasonable rehabilitation plan is formulated.

Description

Cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system

Technical Field

The invention relates to the technical field of cardiovascular and cerebrovascular monitoring, in particular to an intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation.

Background

The technical field of cardiovascular and cerebrovascular monitoring, which is a specialized and technical field, focuses on the use of advanced medical techniques and equipment to monitor, evaluate, and assist in the rehabilitation process of patients suffering from cardiovascular and cerebrovascular diseases. In this field, emphasis is placed not only on diagnosis of diseases but also on tracking of disease progress, evaluation of therapeutic effects, and long-term rehabilitation management. Comprehensive applications of various technologies, such as biosensing technology, data analysis, telemedicine technology, and artificial intelligence algorithms, are often included to enable continuous and accurate monitoring of the cardiovascular and cerebrovascular health status of a patient.

The intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation is a composite system combining modern information technology and medical knowledge, and aims to provide comprehensive rehabilitation support for patients suffering from cardiovascular and cerebrovascular diseases. The system is mainly used for providing real-time data and health indexes for doctors and patients by continuously monitoring the cardiovascular and cerebrovascular health conditions of the patients, so that the treatment scheme and the rehabilitation plan are optimized. In this way, the system aims to improve the rehabilitation effect of the patient, reduce the risk of relapse and promote the overall quality of life.

The traditional cardiovascular and cerebrovascular rehabilitation system has the defects in complex data processing and personalized rehabilitation scheme provision. First, conventional systems often lack efficient data processing and analysis capabilities, resulting in inaccurate or untimely data interpretation, and difficulty in accurately identifying health trends and outliers. In terms of behavior pattern recognition and rehabilitation strategy simulation, the lack of deep learning and advanced data analysis technology makes it difficult to provide personalized analysis and accurate strategy prediction. Insufficient real-time health monitoring results in a potential risk response lag, increasing the risk of the rehabilitation process. In general, traditional systems are limited in terms of data processing efficiency, prediction accuracy, and personalized rehabilitation scheme provision, resulting in poor rehabilitation results and prolonged rehabilitation time.

Disclosure of Invention

The invention aims to solve the defects existing in the prior art, and provides an intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation.

In order to achieve the above purpose, the present invention adopts the following technical scheme: the intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation comprises a physiological data monitoring module, a health state analysis module, a behavior pattern recognition module, a rehabilitation scene simulation module, a real-time data processing module, a heart and brain network analysis module, a rehabilitation progress prediction module and a rehabilitation tracking module;

The physiological data monitoring module processes physiological signals by adopting a Fourier transform algorithm based on data acquired by the sensor, performs signal filtering and denoising, performs normalization processing by a data normalization method, and performs feature extraction on the data to generate standardized physiological data;

the health state analysis module is used for identifying health trends by adopting a time sequence analysis method based on standardized physiological data, carrying out data classification and abnormal point detection by utilizing a statistical method, evaluating health states by utilizing a trend analysis algorithm, and generating health state evaluation;

The behavior pattern recognition module recognizes a behavior pattern by applying an association rule mining method based on standardized physiological data and user activity records, performs pattern classification by using a cluster analysis method, determines health influence factors by statistical analysis, and generates a behavior pattern influence factor analysis result;

The rehabilitation scene simulation module predicts the effect of the differentiated rehabilitation strategy by adopting a Monte Carlo simulation method based on the analysis results of the health state evaluation and the behavior mode influence factors, then evaluates the applicability of the strategy by adopting a decision tree analysis method, and optimizes the strategy selection through effect feedback to generate rehabilitation strategy simulation;

the real-time data processing module is used for carrying out data processing by applying an edge computing technology based on standardized physiological data, carrying out real-time health monitoring by utilizing a stream data analysis technology, mining potential health risks by an anomaly detection algorithm, and generating instant health monitoring feedback;

The heart brain network analysis module analyzes the structural characteristics of the heart brain blood vessel network by adopting a graph theory analysis method based on standardized physiological data, identifies key nodes in the network by adopting a network topology method, predicts potential risks by evaluating network stability, and generates heart brain network analysis results;

The rehabilitation process prediction module is used for constructing a rehabilitation process model by adopting a prediction modeling method based on health state evaluation and heart brain network analysis results, predicting future trends by adopting a time sequence analysis method, adjusting the prediction results by a model optimization technology and generating rehabilitation process prediction results;

The rehabilitation tracking module is used for tracking the rehabilitation effect by utilizing a progress monitoring technology based on the rehabilitation progress prediction result, then quantitatively analyzing by applying an effect evaluation method, optimizing the rehabilitation path by adjusting a scheme, and generating a rehabilitation effect tracking.

As a further scheme of the invention, the standardized physiological data comprises heart rate waveforms, blood pressure values and respiratory frequency, the health state evaluation comprises abnormal value alarms, health indexes and trend prediction, the behavior pattern influence factor analysis results comprise activity habit classification, daily behavior association and health influence score, the rehabilitation strategy simulation comprises strategy effect score, strategy adaptability evaluation and potential risk identification, the instant health monitoring feedback comprises real-time health alarms, instant parameter changes and response suggestions, the heart brain network analysis results comprise network connection diagrams, key node analysis and network stability evaluation, the rehabilitation process prediction results comprise future health state trends, potential risk evaluation and intervention opportunity judgment, and the rehabilitation effect tracking comprises rehabilitation progress tracking, effect quantitative analysis and strategy adjustment suggestion.

As a further scheme of the invention, the physiological data monitoring module comprises a data acquisition sub-module, a signal processing sub-module and a data standardization sub-module;

The data acquisition sub-module is based on a wearable sensor, adopts a photoelectric plethysmography and a vibration type pressure sensing technology, collects original physiological data comprising heart rate and blood pressure, captures the change of heart activity and blood flow fluctuation through the sensor, converts the change into an electric signal and generates the original physiological data;

The signal processing submodule analyzes the frequency domain characteristics of the signals by adopting a fast Fourier transform algorithm based on the original physiological data, removes high-frequency noise by using a low-pass filter, extracts main components of the signals and generates the processed physiological data;

The data normalization submodule calculates the average value and standard deviation of each index by using a Z score normalization method based on the processed physiological data, converts and matches the data with standard normal distribution, and identifies key features related to the health condition of the heart and the brain, including heart rate variability, through feature extraction, so that the universality and comparability of the data are enhanced, and standardized physiological data are generated.

As a further scheme of the invention, the health status analysis module comprises a data analysis sub-module, a status evaluation sub-module and a trend analysis sub-module;

The data analysis submodule carries out dimension reduction processing on the data by adopting a principal component analysis algorithm based on standardized physiological data, simplifies a data structure and reserves key information, then classifies the processed data by using cluster analysis, detects abnormal points in the data by using a support vector machine algorithm, and generates an abnormal point identification and data classification result;

The state evaluation submodule analyzes the time series data by adopting an autoregressive moving average model based on abnormal point identification and data classification results, identifies key trends and modes in the health data, analyzes potential state transitions behind the trends and the modes by utilizing a hidden Markov model, analyzes dynamic changes of the data and generates trend identification and change point analysis results;

The trend analysis submodule carries out trend analysis on the data by adopting linear regression based on trend identification and change point analysis results, determines the key development direction of the health state, and simultaneously smoothes the data sequence by applying a sliding average technology to generate health state assessment.

As a further scheme of the invention, the behavior pattern recognition module comprises an activity data analysis sub-module, a pattern mining sub-module and an influence factor recognition sub-module;

The activity data analysis submodule records daily activities of the user by adopting a time sequence analysis algorithm based on standardized physiological data and user activity records, comprises walking and running, analyzes time distribution characteristics and intensity changes of daily activity data, reveals activity rules of the user, analyzes daily behavior patterns of the user by identifying periodicity and abnormal points of the activity data, and generates an activity data analysis result;

The pattern mining submodule adopts an Apriori algorithm to carry out association rule mining on user behaviors based on the analysis result of the activity data, analyzes association patterns between the user activity data and the physiological data, reveals the relationship between target behaviors and physiological index changes by capturing frequently-occurring behavior combinations and abnormal patterns, mines key behavior patterns affecting the health of the user, and generates behavior association pattern results;

The influence factor identification submodule classifies the behavior patterns by adopting a K-means clustering algorithm based on the behavior association pattern result, reveals the structure and the type of the behavior patterns by dividing the behavior patterns into differentiated categories, evaluates the statistical difference of the differentiated behavior patterns on health by using variance analysis, thereby determining the influence factors of the behavior patterns on health and generating a behavior pattern influence factor analysis result.

As a further scheme of the invention, the rehabilitation scene simulation module comprises a strategy simulation sub-module, an effect evaluation sub-module and a simulation feedback sub-module;

the strategy simulation sub-module is used for predicting the effect of the rehabilitation strategy by adopting a Monte Carlo simulation algorithm based on health state evaluation and behavior pattern analysis, and generating a predicted result of simulating multiple rehabilitation scenes by generating a batch of random samples to generate a predicted result of the effect of the rehabilitation strategy;

The effect evaluation submodule evaluates the applicability of the strategies based on the effect prediction result of the rehabilitation strategies by adopting a decision tree analysis method, evaluates the applicability of the multiple strategies under the differentiated health state and behavior mode by constructing a decision tree to analyze the potential effect of the differentiated strategies, and generates a strategy applicability evaluation result;

the simulation feedback sub-module performs effect feedback optimization based on the strategy applicability evaluation result, adjusts rehabilitation strategy selection, comprises the steps of comparing differentiated rehabilitation strategies and selecting an optimal strategy, and simultaneously generates rehabilitation strategy simulation by referring to the conditions and preferences of patients.

As a further scheme of the invention, the real-time data processing module comprises an edge computing sub-module, a data flow analysis sub-module and a real-time feedback sub-module;

The edge computing sub-module performs data processing on local equipment by adopting an edge computing technology based on standardized physiological data, and performs data processing and response including real-time data receiving, analysis and preliminary processing by dispersing data processing tasks to a plurality of network edge nodes to generate edge processing data;

The data flow analysis submodule analyzes continuous physiological data flow in real time by applying APACHE FLINK flow data processing technology based on edge processing data, monitors the change of heart rate and blood pressure physiological parameters in real time by processing and evaluating each data point, captures the fluctuation of health state in real time and generates real-time data flow analysis results;

The real-time feedback submodule analyzes and identifies abnormal modes in real-time data, including abnormal heart rate fluctuation or blood pressure change, based on real-time data flow analysis results by adopting an abnormal detection algorithm based on statistics and machine learning, and timely identifies potential health risks to generate instant health monitoring feedback.

As a further scheme of the invention, the heart brain network analysis module comprises a network construction sub-module, a structure analysis sub-module and a risk identification sub-module;

the network construction submodule adopts a graph theory analysis algorithm to construct a cardiovascular and cerebrovascular network model by calculating the connectivity and the shortest path between nodes based on standardized physiological data, identifies the basic structure and the connection mode of the network and generates a heart and brain network structure model;

The structure analysis submodule is based on a heart brain network structure model, adopts a network topology analysis method, identifies key nodes in a network through centrality analysis, calculates cluster coefficients to reveal aggregation characteristics among the nodes, and divides a subset group of the network through modularity analysis to generate a network key node analysis result;

the risk identification submodule adopts a dynamic network stability evaluation method based on network key node analysis results, tracks dynamic changes among nodes by time sequence analysis, evaluates the stability of the whole network by a dynamic graph model and generates a heart brain network analysis result.

As a further scheme of the invention, the rehabilitation process prediction module comprises a process building module, a prediction analysis sub-module and a trend prediction sub-module;

The process modeling module analyzes the historical trend of the health data by adopting a linear regression model based on the health state evaluation and the heart brain network analysis result, processes heart brain network characteristics by combining a logistic regression model, further captures key change factors in the rehabilitation process, establishes the integral description of the rehabilitation process and generates a rehabilitation process model;

The prediction analysis submodule is used for carrying out time sequence analysis by utilizing an autoregressive integral moving average model based on a rehabilitation progress model, analyzing historical health data, predicting future health state changes of a patient, identifying future health risks and rehabilitation trends and generating a rehabilitation trend prediction result;

And the trend prediction submodule evaluates the prediction effect of the model by applying a cross verification method based on the rehabilitation trend prediction result, optimizes model parameters by using a grid search technology and generates a rehabilitation process prediction result.

As a further scheme of the invention, the rehabilitation tracking module comprises an effect monitoring sub-module, a progress evaluation sub-module and an adjustment suggestion sub-module;

the effect monitoring submodule is used for monitoring physiological parameters of a patient in real time, including heart rate and muscle activity, based on a rehabilitation progress prediction result by adopting heart rate variability analysis and an electromyography signal processing algorithm, evaluating the instant change of a rehabilitation effect and generating rehabilitation state real-time monitoring data;

The progress evaluation sub-module is used for tracking the rehabilitation progress by adopting time sequence analysis based on the rehabilitation state real-time monitoring data, evaluating the rehabilitation speed and efficiency by combining variance analysis and regression analysis, performing multidimensional quantitative analysis of the rehabilitation process, and generating comprehensive rehabilitation progress evaluation;

the adjustment suggestion submodule analyzes the trend of rehabilitation data by adopting a machine learning prediction model based on comprehensive assessment of the rehabilitation progress, adjusts a rehabilitation scheme according to the current state of a patient by combining a path planning algorithm, provides personalized rehabilitation suggestions and generates rehabilitation effect tracking.

Compared with the prior art, the invention has the advantages and positive effects that:

In the invention, physiological signals are processed through a Fourier transform algorithm, so that the data quality is improved, the accuracy of health trend identification and abnormal point detection is enhanced by combining a time sequence analysis method with a statistical method, and deep insight is provided in behavior pattern identification by association rule mining and cluster analysis, so that the identification of health influence factors is more comprehensive. The Monte Carlo simulation and decision tree analysis method optimizes the recovery strategy selection, and improves the recovery effect. And the edge calculation and stream data analysis technology is applied to realize efficient real-time health monitoring and timely discover potential risks. The graph theory analysis and network topology method accurately predicts potential risks in the heart-brain network analysis, and improves the pertinence of preventive measures. Prediction modeling and time series analysis provide accurate future trend prediction in rehabilitation process prediction, and assist reasonable rehabilitation planning.

Drawings

FIG. 1 is a system flow diagram of the present invention;

FIG. 2 is a schematic diagram of a system framework of the present invention;

FIG. 3 is a flow chart of a physiological data monitoring module according to the present invention;

FIG. 4 is a flow chart of a health status analysis module according to the present invention;

FIG. 5 is a flow chart of a behavior pattern recognition module according to the present invention;

FIG. 6 is a flow chart of a rehabilitation scenario simulation module of the present invention;

FIG. 7 is a flow chart of a real-time data processing module of the present invention;

FIG. 8 is a flowchart of a heart brain network analysis module according to the present invention;

FIG. 9 is a flow chart of a rehabilitation process prediction module according to the present invention;

FIG. 10 is a flow chart of a rehabilitation tracking module according to the present invention.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

In the description of the present invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like indicate orientations or positional relationships based on the orientation or positional relationships shown in the drawings, merely to facilitate describing the present invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present invention. Furthermore, in the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.

Embodiment one: referring to fig. 1 to 2, an intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation includes a physiological data monitoring module, a health status analysis module, a behavior pattern recognition module, a rehabilitation scene simulation module, a real-time data processing module, a heart brain network analysis module, a rehabilitation progress prediction module, and a rehabilitation tracking module;

The physiological data monitoring module processes physiological signals by adopting a Fourier transform algorithm based on data acquired by the sensor, performs signal filtering and denoising, performs normalization processing by a data normalization method, and performs feature extraction on the data to generate normalized physiological data;

The health state analysis module is used for identifying health trends based on standardized physiological data by adopting a time sequence analysis method, carrying out data classification and abnormal point detection by utilizing a statistical method, evaluating health states by utilizing a trend analysis algorithm, and generating health state evaluation;

the rehabilitation scene simulation module predicts the effect of the differentiated rehabilitation strategy by adopting a Monte Carlo simulation method based on the health state evaluation and the behavior mode influence factor analysis result, then evaluates the applicability of the strategy by adopting a decision tree analysis method, optimizes the strategy selection through effect feedback, and generates rehabilitation strategy simulation;

The heart-brain network analysis module analyzes the structural characteristics of the heart-brain blood vessel network by adopting a graph theory analysis method based on standardized physiological data, identifies key nodes in the network by adopting a network topology method, predicts potential risks by evaluating network stability, and generates heart-brain network analysis results;

the rehabilitation process prediction module is used for constructing a rehabilitation process model by adopting a prediction modeling method based on health state evaluation and heart brain network analysis results, predicting future trends by adopting a time sequence analysis method, adjusting the prediction results by adopting a model optimization technology, and generating a rehabilitation process prediction result;

The standardized physiological data comprise heart rate waveforms, blood pressure values and respiratory frequency, the health state assessment comprises abnormal value alarms, health indexes and trend prediction, the behavior mode influence factor analysis results comprise activity habit classification, daily behavior association and health influence score, the rehabilitation strategy simulation comprises strategy effect score, strategy adaptability evaluation and potential risk identification, the instant health monitoring feedback comprises real-time health alarms, instant parameter changes and response suggestions, the heart brain network analysis results comprise network connection diagrams, key node analysis and network stability assessment, the rehabilitation progress prediction results comprise future health state trends, potential risk assessment and intervention opportunity judgment, and the rehabilitation effect tracking comprises rehabilitation progress tracking, effect quantitative analysis and strategy adjustment suggestions.

In a physiological data monitoring module, by using sensor-based data collection, the data is first subjected to signal processing via a fourier transform algorithm, steps involving converting time series data into the frequency domain, in order to better understand and analyze the periodic components in the signal. Next, digital signal processing techniques are used to filter and denoise the signals, in which a bandpass filter is used primarily to remove high and low frequency noise from the signals, leaving an effective portion of the physiological signal. The data is then normalized by a normalization process, which typically involves converting the raw data into a form with zero mean and unit variance, thereby making the data from different sources comparable. Next, the feature extraction step may extract key features in the data through an algorithm, such as Principal Component Analysis (PCA) or Linear Discriminant Analysis (LDA), to facilitate subsequent analysis. Through these careful processing steps, standardized physiological data are ultimately generated, which are not only uniform in format, but also improved in quality, providing a reliable basis for subsequent health status analysis.

In the health status analysis module, based on standardized physiological data, a time series analysis method is first applied to identify health trends, which involves analyzing time dependencies in the data using an autoregressive moving average (ARMA) model or an autoregressive integral moving average (ARMA) model, thereby revealing the law of change of health status over time. Then, data classification and outlier detection are performed using statistical methods, such as k-means clustering or Support Vector Machines (SVMs), to distinguish between normal and outlier health states. Then, the current health status is evaluated by a trend analysis algorithm such as linear regression or polynomial regression, and a future development trend is predicted. The combination of these steps not only provides insight into the health of the individual, but also predicts future health risks, and the resulting health assessment report provides valuable reference information for the user and medical professionals.

In the behavior pattern recognition module, based on standardized physiological data and user activity records, a correlation rule mining method is firstly adopted to recognize a behavior pattern, and the step is mainly to analyze the correlation between data items through an algorithm such as Apriori or FP-Growth so as to reveal the internal relation between different physiological indexes and user behaviors. The behavior patterns are then classified using a cluster analysis method, such as spectral clustering or hierarchical clustering, to determine the categories and features of the patterns. Furthermore, the application of statistical analysis helps to determine which patterns of behavior have significant associations with health status. Finally, the behavior pattern generated by the module influences the factor analysis result, so that personalized health advice can be provided, and better health management in daily life can be assisted.

In the rehabilitation scene simulation module, based on the health state evaluation and the behavior mode influence factor analysis result, different rehabilitation strategy effects are simulated by adopting a Monte Carlo simulation method. By means of a random sampling technology, the method can generate a large number of rehabilitation scenes and evaluate potential effects of each strategy under different conditions. Then, the applicability of different rehabilitation strategies is evaluated by applying a decision tree analysis method, the steps relate to constructing a decision tree model, and the optimal rehabilitation strategy is determined by analyzing the influence of different characteristics on the rehabilitation effect. Finally, the strategy selection is optimized by effect feedback, namely data in the rehabilitation process is continuously collected and analyzed, and the series of simulation and evaluation processes can not only provide personalized and accurate rehabilitation schemes, but also improve the success rate of rehabilitation through continuous optimization processes.

In a real-time data processing module, rapid data processing is performed by applying an edge computing technique that allows data processing to be performed on devices closer to the user, thereby reducing latency and improving processing efficiency. Next, real-time health monitoring is performed using flow data analysis techniques. This includes processing and analyzing real-time data streams using frameworks such as SPARK STREAMING or Flink to discover changes in health trends in time. Anomaly detection algorithms, such as isolated forests or neural networks, are used to mine potential health risks. The algorithms can quickly identify abnormal modes from real-time data and help users to find health problems in time. The generated instant health monitoring feedback not only improves the real-time performance and accuracy of health management, but also provides timely health warning for users.

In the heart brain network analysis module, based on standardized physiological data, a graph theory analysis method is firstly adopted to analyze the structural characteristics of a heart brain blood vessel network, in the process, the data is regarded as a network, wherein nodes represent different physiological parameters, and edges represent correlations among the parameters. The overall structure and local connection mode of the network can be revealed by calculating various indexes of the network, such as degree distribution, clustering coefficients and the like. Network topology methods are then applied to identify key nodes in the network that play an important role in maintaining overall network stability. Furthermore, potential health risks can be predicted by evaluating the stability of the heart brain network, such as the robustness or vulnerability of the computing network. The heart and brain network analysis result generated by the module provides a new view angle for understanding the heart and brain health condition of the individual and is helpful for preventing and managing related diseases.

In the rehabilitation process prediction module, based on the health state evaluation and the analysis result of the heart and brain network, a prediction modeling method is firstly adopted to construct a rehabilitation process model, and the process may involve using a machine learning algorithm, such as a random forest or gradient elevator, to analyze past rehabilitation data and current health states so as to predict future trend of rehabilitation. Time series analysis is then applied to improve the accuracy of the predictions, for example by adjusting time windows or seasonal components in the model to accommodate the characteristics of the rehabilitation data. Finally, model parameters are adjusted through model optimization technology such as cross validation or grid search, and accuracy and reliability of a prediction result are ensured. The generated rehabilitation process prediction result not only helps individuals and medical professionals to better understand the rehabilitation process, but also provides scientific basis for making and adjusting the rehabilitation plan, and ensures the high efficiency and individuation of the rehabilitation path.

In the rehabilitation tracking module, based on the rehabilitation progress prediction result, the rehabilitation effect is tracked by utilizing a progress monitoring technology. This includes periodically collecting and analyzing key indicators during rehabilitation, such as physical activity, physiological data changes, etc., to monitor rehabilitation progress and effectiveness. Quantitative analysis is then performed by applying an effect assessment method, which involves using statistical tests and comparative analysis to assess the effectiveness of rehabilitation measures. For example, the physiological index changes before and after rehabilitation are compared, or regression models are used to analyze the influence of different factors on rehabilitation effect. In addition, the rehabilitation path is optimized by adjusting the protocol, and the steps include adjusting the rehabilitation plan based on the result of the effect evaluation, such as changing the intensity or frequency of rehabilitation training, or introducing new rehabilitation measures. The generated recovery effect tracking report not only provides clear recovery targets and guidance for individuals, but also is beneficial to timely adjusting and optimizing recovery schemes and ensures the continuity and effectiveness of recovery processes.

Referring to fig. 3, the physiological data monitoring module includes a data acquisition sub-module, a signal processing sub-module, and a data standardization sub-module;

The data acquisition submodule is based on a wearable sensor, adopts photoelectric plethysmography and vibration pressure sensing technology, collects original physiological data comprising heart rate and blood pressure, captures the change of heart activity and blood flow fluctuation through the sensor, converts the change into an electric signal and generates the original physiological data;

The signal processing sub-module analyzes the frequency domain characteristics of the signals by adopting a fast Fourier transform algorithm based on the original physiological data, removes high-frequency noise by using a low-pass filter, extracts the main components of the signals and generates the processed physiological data;

the data standardization submodule calculates the average value and standard deviation of each index by using a Z score standardization method based on the processed physiological data, converts and matches the data with standard normal distribution, and identifies key characteristics related to heart and brain health conditions, including heart rate variability, through characteristic extraction, so that the universality and comparability of the data are enhanced, and standardized physiological data are generated.

In the data acquisition sub-module, the photoelectric plethysmography and the vibration type pressure sensing technology are adopted through the wearable sensor, so that the fine changes of the heart activity and the blood flow fluctuation are effectively captured. Photoplethysmography relies on the transmission and reception of reflected light to measure changes in blood volume, and sensors monitor heart rate by analyzing optical changes in blood volume under the skin. Specifically, when the heart beats, blood flow causes changes in the intensity of light reflected under the skin, which are captured by the sensor and converted into electrical signals. These electrical signals form raw heart rate data, typically expressed in beats per minute (bpm). The vibration pressure sensing technology measures blood pressure by sensing minute vibrations of the skin surface due to blood flow. This technique utilizes piezoelectric materials or microelectromechanical system (MEMS) sensors to detect blood pressure fluctuations, and converts these physical changes into electrical signals, thereby generating raw blood pressure data, typically in millimeters of mercury (mmHg). These raw physiological data, stored in time series, including heart rate and blood pressure readings at each time point, provide the basis for subsequent signal processing and analysis.

In the signal processing sub-module, the raw physiological data is processed through a Fast Fourier Transform (FFT) algorithm to analyze the frequency domain characteristics of the signal. The FFT algorithm converts the signal in the time domain into a frequency domain representation so that the intensities of the different frequency components in the signal can be analyzed. In this process, the raw data is first windowed to reduce edge effects. Then, applying an FFT algorithm to convert the data from the time domain to the frequency domain, the steps reveal the magnitudes of the individual frequency components in the heart rate and blood pressure signals, helping to identify and separate the primary physiological signal and background noise. Then, the high frequency noise is removed by a low pass filter, leaving the main components of the signal. The low-pass filter sets a specific cut-off frequency, only allows the signal component lower than the cut-off frequency to pass through, so that high-frequency noise is removed, and after the step, physiological data subjected to noise removal and signal enhancement are generated, and the physiological data more accurately reflect the real conditions of the heart and blood vessels.

In the data normalization sub-module, a Z-score normalization method is applied to the processed physiological data to improve the universality and comparability of the data. Z-score normalization is a method of converting data into a form with zero mean and unit standard deviation, so that data from different sources or of different magnitudes can be fairly compared. In a specific operation, the average and standard deviation of each index (such as heart rate and blood pressure) are first calculated. Then, each data point is subtracted from the average value of the corresponding index and divided by the standard deviation to produce a Z-score, and the conversion process redistributes all data points over a standard normal distribution, wherein the value of each point represents the degree of deviation thereof from the average level. Further, key features, such as heart rate variability, that are closely related to heart and brain health conditions are extracted from these normalized data by feature extraction methods, such as Principal Component Analysis (PCA). These features not only reflect the health of the heart, but also reveal potential links between heart and brain health. Finally, the generated standardized physiological data is optimized in format and quality, and a solid foundation is provided for deeper health analysis and model construction.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, a hypertensive patient is taken as an example. Raw data includes heart rate per minute (e.g., 75 bpm) and blood pressure readings (e.g., 140/90 mmHg). These data are first collected and recorded by the sensor technology of the data acquisition sub-module. In the signal processing sub-module, an FFT algorithm is applied to convert these data into a frequency domain representation, analyzing the main frequency components of the heart rate and blood pressure signals. Then, the high frequency noise is removed by a low pass filter, and clearer heart and blood vessel activity signals are obtained. Next, in the data normalization sub-module, these processed data are Z-score normalized, such as converting heart rate 75 bpm to a Z-score relative to the average heart rate, and key features related to heart brain health, such as heart rate variability, are identified by feature extraction. Therefore, the finally generated standardized physiological data is unified in format and is remarkably improved in quality, and scientific basis is provided for subsequent health state evaluation and rehabilitation scheme formulation.

Referring to fig. 4, the health status analysis module includes a data analysis sub-module, a status evaluation sub-module, and a trend analysis sub-module;

The data analysis sub-module adopts a principal component analysis algorithm to carry out dimension reduction processing on the data based on standardized physiological data, simplifies the data structure and retains key information, then uses cluster analysis to classify the processed data, and detects abnormal points in the data through a support vector machine algorithm to generate abnormal point identification and data classification results;

The state evaluation submodule analyzes the time series data by adopting an autoregressive moving average model based on abnormal point identification and data classification results, identifies key trends and modes in the health data, analyzes potential state transitions behind the trends and modes by utilizing a hidden Markov model, analyzes dynamic changes of the data and generates trend identification and change point analysis results;

the trend analysis sub-module adopts linear regression to carry out trend analysis on the data based on the trend identification and the change point analysis result, determines the key development direction of the health state, and simultaneously adopts a sliding average technology to smooth the data sequence to generate the health state assessment.

In the data analysis submodule, based on standardized physiological data, the data is subjected to dimension reduction processing through a Principal Component Analysis (PCA) algorithm, and the covariance matrix of the data set is calculated firstly in the process, so that the eigenvalues and eigenvectors of the covariance matrix are extracted. The eigenvectors with larger eigenvalues represent the dominant directions of variation in the data, with the data in these directions varying most significantly. The PCA projects the multidimensional data into a space with lower dimensionality by selecting a plurality of feature vectors corresponding to the maximum feature values, so that the purpose of reducing the dimensionality of the data is realized. For example, the original data contains multiple dimensions such as heart rate, blood pressure, heart rate variability and the like, and only two or three main components are reserved after PCA processing, and the components comprehensively reflect the main information of the original data, but the data volume is smaller and the processing is easier. Next, the reduced-dimension data is classified using a cluster analysis, such as a K-means clustering algorithm. The algorithm firstly randomly selects K points as initial cluster centers, then distributes each data point to the nearest cluster center, recalculates the center of each cluster, and repeats the process until the cluster center is stable. In this way, similar data points are classified into the same class, revealing the inherent distribution rules in the data. In addition, support Vector Machine (SVM) algorithms are used to detect outliers in the data. The SVM separates the different categories of data points by constructing a hyperplane, and outlier detection is the process of identifying outliers that are far from most of the data points. These operations generate outlier recognition and data classification results, providing accurate underlying data for subsequent state assessment.

In the state evaluation sub-module, based on the outlier identification and the data classification result, the time series data is first analyzed by adopting an autoregressive moving average (ARMA) model. ARMA model combines two methods, autoregressive (AR) and Moving Average (MA), to effectively identify key trends and patterns in health data. The autoregressive portion reveals the relationship between the current data point and its historical value, while the moving average portion smoothes random fluctuations in the time series. For example, by analyzing the time series of heart rate data, the ARMA model may identify a steady trend or periodic fluctuation of heart rate over time. The potential state transitions behind these trends and patterns are then analyzed using Hidden Markov Models (HMMs). HMM is a statistical model that assumes that the state of the system is not directly visible, but that changes in state can be inferred from observed data sequences. By HMM, dynamic changes behind such health data as heart rate increases or decreases, such as increases or decreases in heart load, can be resolved. These analysis results generate trend recognition and change point analysis results, providing insight into understanding and assessing the health status of individuals.

In the trend analysis submodule, based on the trend identification and the change point analysis result, linear regression is adopted to carry out trend analysis on the data. Linear regression is a predictive analysis technique that reveals the trend of data by constructing a linear relationship between data points and a predicted target. For example, by linear regression analysis of long-term data of heart rate and blood pressure, it is possible to determine the key direction of development of these indicators over time, such as whether a steady upward or downward trend is exhibited. In addition, a sliding average technique is applied to smooth the data sequence, reducing the impact of short term fluctuations on trend analysis. The moving average generates a smooth data curve by calculating the average value of each point and surrounding points in the data sequence, which is helpful for clearly revealing the long-term trend of the health data, and the series of analysis generates health state assessment, which not only displays the overall trend of the health condition of an individual, but also provides basis for formulating preventive measures or intervention schemes.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, it is assumed that standardized physiological data of a patient include indexes such as heart rate, blood pressure, heart rate variability and the like. In the data analysis sub-module, the multidimensional data are converted into several components containing main change information through PCA dimension reduction processing. The data is then classified into several categories using K-means clustering, e.g., data points with higher and lower heart rates are classified into different categories. The SVM algorithm identifies data points with abnormally high or low heart rates. In the state assessment sub-module, the ARMA model analyzes the time series of heart rate data, identifies steady trends or periodic fluctuations, and then the HMM reveals heart state transitions behind these changes, such as transitions from calm to pressure states. In the trend analysis sub-module, linear regression reveals the main development direction of heart rate and blood pressure over time, and the sliding average technique smoothes out these data, highlighting the long-term trend. Together, these analysis results constitute an assessment of the health status of the patient, providing important data support for their rehabilitation and health management.

Referring to fig. 5, the behavior pattern recognition module includes an activity data analysis sub-module, a pattern mining sub-module, and an influence factor recognition sub-module;

The activity data analysis submodule is used for recording daily activities of the user by adopting a time sequence analysis algorithm based on standardized physiological data and user activity records, analyzing time distribution characteristics and intensity changes of daily activity data, revealing activity rules of the user, and generating an activity data analysis result by identifying periodicity and abnormal points of the activity data and analyzing daily behavior patterns of the user;

The pattern mining submodule is used for carrying out association rule mining on user behaviors by adopting an Apriori algorithm based on the analysis result of the activity data, analyzing association patterns between the user activity data and the physiological data, revealing the relationship between target behaviors and physiological index changes by capturing frequently-occurring behavior combinations and abnormal patterns, mining key behavior patterns affecting the health of the user, and generating behavior association pattern results;

The influence factor identification submodule classifies the behavior patterns based on the behavior association pattern result by adopting a K-means clustering algorithm, reveals the structure and the type of the behavior patterns by dividing the behavior patterns into differentiated categories, evaluates the statistical difference of the differentiated behavior patterns on health by using variance analysis, determines the influence factors of the behavior patterns on health, and generates a behavior pattern influence factor analysis result.

In the activity data analysis sub-module, based on the standardized physiological data and the user activity records, a time series analysis algorithm is adopted to record the daily activities of the user. The data formats include time stamps, type of activity (e.g., walking, running), duration and intensity of activity, etc. The time series analysis algorithm first pre-processes the data, including removing outliers and filling in missing data. The algorithm then analyzes the time distribution characteristics and intensity variations of the daily activity data. For example, moving average or exponential smoothing methods are used to smooth the activity intensity data revealing long-term trends and seasonal patterns of user activity. Next, by identifying the periodicity and outliers of the activity data, the algorithm can analyze the daily behavior patterns of the user. The periodic analysis involves an autocorrelation function (ACF) and a partial autocorrelation function (PACF) to identify repetitive patterns in the activity data, such as movement habits at specific times per week. Outlier detection then uses statistical methods, such as Z-score or box-plot methods, to identify those data points that differ significantly from the regular activity pattern. The detailed analysis operation generates an activity data analysis result, reveals the activity rule and abnormal behavior of the user, and provides a basis for subsequent pattern mining and influence factor identification.

In the pattern mining sub-module, based on the analysis result of the activity data, the Apriori algorithm is adopted to carry out association rule mining on the user behavior, and in the process, firstly, the user behavior and the physiological data are converted into a format suitable for association rule mining, for example, the activity type and the heart rate threshold value are combined into a term set. The Apriori algorithm then iteratively generates frequent term sets, i.e., combinations of behaviors that occur frequently together in the data set, which includes calculating the frequency with which the term sets occur in the data, and pruning out those term sets that are below a preset support threshold. The algorithm then derives association rules from the set of frequent items that represent potential links between specific behavior patterns and physiological data changes, such as a relationship between specific exercise intensity and heart rate rise. By capturing the frequently-occurring behavior combinations and abnormal patterns, the Apriori algorithm reveals the relationship between target behaviors and physiological index changes, so that key behavior patterns affecting the health of users are mined, and behavior association pattern results are generated.

And in the influence factor identification sub-module, classifying the behavior patterns by adopting a K-means clustering algorithm based on the behavior association pattern result. The method comprises the steps of firstly determining the number K of clusters, then iteratively distributing the behavior patterns into different clusters through an algorithm, and finally, each cluster represents a differentiated behavior pattern category. Statistical differences in the health effects of behavior patterns in the different clusters were then evaluated using analysis of variance (ANOVA). This involves calculating the average difference inside and between clusters to determine the significance of different behavioral patterns on health effects. Completion of these steps helps identify the behavior pattern that has the greatest impact on the user's health, generating behavior pattern impact factor analysis results. These results not only reveal the close relationship between the user's health status and the daily behavior patterns, but also provide an important basis for developing personalized health improvement plans.

In the cardiovascular and cerebrovascular rehabilitation intelligent assistance system, it is assumed that the standardized physiological data of the user includes daily heart rate and blood pressure recordings, and the activity recordings include daily walking and running durations and intensities. In the activity data analysis sub-module, the time series analysis algorithm first smoothes the activity intensity data revealing the increasing trend of the user's weekly running activity and longer weekend walking activity. Occasional high intensity running activities are then identified by outlier detection. In the pattern mining sub-module, the Apriori algorithm finds an associated pattern in which the heart rate is significantly increased after the user runs at high intensity. Finally, in the influence factor identification submodule, the K-means clustering algorithm divides the user behaviors into different categories, such as regular walking and occasional high-intensity running, and the variance analysis reveals the influence degree of different behavior modes on heart rate and blood pressure, so that a behavior mode influence factor analysis result is generated. These analysis results provide specific advice to the user on how to adjust daily activities to improve cardiovascular and cerebrovascular health.

Referring to fig. 6, the rehabilitation scenario simulation module includes a strategy simulation sub-module, an effect evaluation sub-module, and a simulation feedback sub-module;

The effect evaluation submodule evaluates the applicability of the strategies based on the rehabilitation strategy effect prediction result by adopting a decision tree analysis method, analyzes the potential effect of the differentiated strategies by constructing a decision tree, evaluates the applicability of the multiple strategies under the differentiated health states and behavior modes, and generates a strategy applicability evaluation result;

the simulation feedback sub-module performs effect feedback optimization based on the strategy applicability evaluation result, adjusts rehabilitation strategy selection, comprises comparing differentiated rehabilitation strategies and selecting an optimal strategy, and simultaneously generates rehabilitation strategy simulation by referring to the condition and preference of a patient.

And in the strategy simulation sub-module, based on the health state evaluation and the behavior pattern analysis result, performing effect prediction of the rehabilitation strategy by adopting a Monte Carlo simulation algorithm. Monte Carlo simulation is a random sampling based calculation method for simulating a system with uncertainty. In the rehabilitation strategy simulation, the algorithm first defines parameters of the rehabilitation strategy, such as movement frequency, intensity, type and the like, and health indexes such as heart rate, blood pressure, heart rate variability and the like, which are influenced by the parameters. The algorithm then generates a large number of random samples, each representing a particular rehabilitation strategy configuration. For each sample, the algorithm calculates the expected health effects of the corresponding rehabilitation strategy, such as the magnitude of heart rate reduction, the effect of blood pressure control, etc., according to a predefined model. These calculations are typically based on statistical distributions and historical data to simulate the effects of different rehabilitation strategies in a variety of health states and behavioral patterns. By the method, monte Carlo simulation generates a series of estimated results for each rehabilitation strategy, and potential effects and risks of different strategies are revealed. The finally generated recovery strategy effect prediction result not only shows the expected effects of various strategies under different conditions, but also provides an important basis for selecting the most suitable recovery strategy.

And in the effect evaluation sub-module, based on the rehabilitation strategy effect prediction result, evaluating the applicability of the strategy by adopting a decision tree analysis method. The decision tree is a tree structure model for classification and regression, and the data is divided into different subsets through a series of rules, so that the purpose of decision is finally achieved. In the module, a decision tree is constructed according to the rehabilitation strategy effect prediction result and the health state and behavior mode of the user. For example, one node decides on branches based on the user's blood pressure level, and another node selects different strategies according to the user's exercise habits. The decision tree identifies the rehabilitation strategy most suitable for the specific user group by analyzing the potential effects of different strategies under various health states and behavior modes. The generated strategy applicability evaluation result not only provides decision support for medical professionals, but also provides personalized rehabilitation strategy selection basis for users.

In the simulation feedback sub-module, based on the strategy applicability evaluation result, effect feedback optimization is performed, rehabilitation strategy selection is adjusted, the process involves comparing effects and applicability of different rehabilitation strategies, and an optimal strategy is selected. For example, by comparing the impact of different exercise programs on cardiac rehabilitation, the most effective regimen is selected. At the same time, personal conditions and preferences of the patient are considered, as some patients are more prone to low intensity but frequent movements. These considerations are integrated into the simulated feedback flow to ensure that the selected rehabilitation strategy is not only theoretically effective, but also meets the actual needs and lifestyle of the patient. The finally generated rehabilitation strategy simulation result provides a customized rehabilitation plan for the patient, aims at improving the rehabilitation effect and simultaneously ensures the comfort level and the compliance of the patient.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, the behavior pattern analysis shows that the daily activities of a patient are less on the assumption that the health state evaluation result of the patient shows the problems of hypertension and unstable heart rate. In the strategy simulation sub-module, the Monte Carlo simulation algorithm simulates a variety of rehabilitation strategies, such as exercise plans of different intensities and frequencies, predicting the potential impact of these strategies on the patient's blood pressure and heart rate. In the effect evaluation sub-module, decision tree analysis determines an optimal exercise program for the patient's health condition, such as moderate-intensity aerobic exercise combined with daily walking. Finally, in the simulation feedback sub-module, the rehabilitation strategy is adjusted according to the preference and living habit of the patient, and the optimal rehabilitation plan is selected. The rehabilitation strategy simulation result generated by the process provides a specific and personalized rehabilitation scheme for the patient, and aims at improving the cardiovascular and cerebrovascular health condition of the patient.

Referring to fig. 7, the real-time data processing module includes an edge computing sub-module, a data flow analysis sub-module, and a real-time feedback sub-module;

The edge computing sub-module performs data processing on the local equipment by adopting an edge computing technology based on standardized physiological data, and performs data processing and response including real-time data receiving, analysis and preliminary processing by dispersing data processing tasks to a plurality of network edge nodes to generate edge processing data;

The data flow analysis submodule is used for carrying out real-time analysis on continuous physiological data flow by utilizing APACHE FLINK flow data processing technology based on edge processing data, and real-time monitoring the change of heart rate and blood pressure physiological parameters and capturing the fluctuation of health state in real time through the processing and evaluation of each data point to generate a real-time data flow analysis result;

The real-time feedback submodule analyzes and identifies abnormal modes in real-time data, including abnormal heart rate fluctuation or blood pressure change, by adopting an abnormal detection algorithm based on statistics and machine learning based on a real-time data flow analysis result, and timely identifies potential health risks to generate instant health monitoring feedback.

In the edge calculation sub-module, based on standardized physiological data, the edge calculation technology is adopted to perform data processing on local equipment, so that the decentralization of data processing is realized. The core of the edge computing technology is to disperse the data processing tasks from the cloud to the edge of the network, i.e. close to the data source, such as the user's smart watch, cell phone or home gateway. This approach can significantly reduce the time and bandwidth required for data transmission and increase the response speed. In this process, primary physiological data (e.g., heart rate, blood pressure, etc.) received from the wearable device is first subjected to preliminary processing, including formatting, cleansing (removing noise or erroneous readings), and preliminary analysis (e.g., detecting substantial fluctuations in heart rate) of the data. The data thus processed, referred to as edge-processed data, is stored in the local device in an optimized format ready for further real-time analysis. The method not only accelerates the data processing speed, but also improves the privacy of the data, because sensitive physiological data does not need to be transmitted to a remote server.

In the data stream analysis sub-module, based on the edge processing data, a APACHE FLINK stream data processing technique is applied to analyze the continuous physiological data stream in real time. APACHE FLINK is a high-performance, scalable stream processing framework capable of handling high-throughput data streams. In this module, a data stream processing pipeline is first set up, which is capable of continuously receiving data streams from edge computing nodes. Subsequently, the Flink processes and evaluates each data point in real time, including applying a time window function to monitor changes in physiological parameters such as heart rate and blood pressure over a specific period of time, and applying a data analysis algorithm to capture fluctuations in health status in real time. For example, by calculating the average heart rate and blood pressure over a sliding time window, the Flink can monitor the trend of these parameters in real time. The generated real-time data flow analysis results provide continuous health monitoring for users, timely capture of any abnormal changes in health state, and provide possibility for timely medical intervention.

In the real-time feedback sub-module, based on the real-time data stream analysis result, an anomaly detection algorithm based on statistics and machine learning is adopted to analyze and identify an anomaly mode in the real-time data. These algorithms include statistical analysis methods such as standard deviation analysis, box plot analysis, and machine learning algorithms such as isolated forests, support vector machines, and the like. First, the algorithm evaluates the results of the real-time data stream analysis to identify significant deviations from normal health conditions, such as abnormal heart rate fluctuations or blood pressure changes. Then, in combination with the historical health data of the user and the general health index standard, the machine learning algorithm further learns and identifies potential abnormal modes, and the process enables the algorithm to timely identify potential health risks and generate instant health monitoring feedback. The real-time feedback is important for early detection of health problems, reminding a user to take corresponding measures or seeking medical help, and is beneficial to improving the effect of chronic disease management and reducing the occurrence of acute events.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, a patient is assumed to wear a smart watch, and the device monitors and collects heart rate and blood pressure data of the patient in real time. In the edge calculation sub-module, the intelligent watch performs preliminary processing and formatting on the collected physiological data to generate edge processing data. Then, in the data flow analysis sub-module APACHE FLINK analyzes the data in real time, monitors the fluctuation of heart rate and blood pressure, and captures any abnormal health state changes in real time. Finally, in the real-time feedback sub-module, by applying an abnormality detection algorithm, the system timely recognizes abnormal fluctuation of the heart rate of the patient, generates instant health monitoring feedback, and prompts the patient to need medical intervention through the intelligent watch. Such a system not only enhances the management of chronic diseases, but also increases the speed of coping with sudden health events.

Referring to fig. 8, the heart brain network analysis module includes a network construction sub-module, a structure analysis sub-module, and a risk identification sub-module;

The network construction submodule adopts a graph theory analysis algorithm to construct a cardiovascular and cerebrovascular network model by calculating the connectivity and the shortest path between nodes based on standardized physiological data, identifies the basic structure and the connection mode of the network and generates a cardiovascular and cerebrovascular network structure model;

the risk identification submodule adopts a dynamic network stability evaluation method based on the network key node analysis result, tracks dynamic changes among nodes by utilizing time sequence analysis, evaluates the stability of the whole network by using a dynamic graph model and generates a heart brain network analysis result.

In the network construction submodule, a cardiovascular and cerebrovascular network model is constructed by adopting a graph theory analysis algorithm based on standardized physiological data. Graph analysis in this application aims to abstract complex interactions and relationships of the cardiovascular and cerebrovascular system into mathematical models for better understanding and analysis of their structural characteristics. First, the individual components of the cardiovascular and cerebrovascular system (e.g., ventricles, arteries, brain regions, etc.) are defined as nodes in the network. The connections between nodes, i.e. the edges of the network, are then determined from physiological data, such as blood flow velocity, oxygen saturation, etc. These connections may represent blood flow paths, neural signaling, or other physiological functional associations. Next, the basic structure and connection mode of the network are revealed by calculating the connectivity between nodes (i.e., the number of node connections) and the shortest path (i.e., the shortest connection path between two nodes). For example, a node of high connectivity may represent a major hub in a cardiovascular and cerebrovascular network, while shortest path analysis helps to understand how the signal or blood flow propagates fastest in a cardiovascular event. The generated heart brain network structure model provides an intuitive framework for understanding the complex dynamics of the heart brain vascular system, and is helpful for identifying key components and potential weak links in the network.

In the structure analysis submodule, based on the heart brain network structure model, the network topology analysis method is adopted to deeply analyze the characteristics of the network. Centrality analysis is used to identify key nodes in the network that play an important role in the cardiovascular and cerebrovascular network. For example, centrality (DEGREE CENTRALITY) calculates the number of direct connections per node, revealing the most active or important nodes; the proximity centrality (closeness centrality) evaluates the average distance of a node to all other nodes, reflecting the reachability of the node in the network. The cluster coefficient calculation is used for revealing the aggregation characteristic among the nodes and displaying the formation condition of small groups in the network, which is very useful in understanding the local interaction mode in the heart brain system. Modularity analysis further helps to partition sub-clusters of the network, revealing functionally or structurally independent modules in the cardiac and cerebral system. Through the analysis, the generated analysis results of the key nodes of the network not only reveal the structural characteristics of the heart-brain network, but also are helpful for understanding the functions and potential vulnerability of the network.

In the risk identification sub-module, based on the analysis result of the network key nodes, the stability of the heart and brain network is analyzed by adopting a dynamic network stability assessment method, and the process involves tracking dynamic changes among nodes by using time sequence analysis, for example, by analyzing the interactive changes between heart nodes and brain nodes within a period of time, the dynamic adjustment process in the heart and brain vascular system can be captured. The dynamic graph model is a key tool for evaluating the stability of the whole network, and not only considers the state of the network at a certain moment, but also focuses on the change trend of the state of the network along with time. This approach may reveal how the network adapts to changes in response to internal or external disturbances, such as sudden changes in blood pressure or heart rate anomalies. The generated heart brain network analysis results not only provide quantitative assessment of network stability, but also help identify potential risk factors leading to cardiovascular and cerebrovascular diseases.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, it is assumed that standardized physiological data of a patient include detailed heart rate, blood pressure and brain activity information. In the network construction sub-module, these data are used to construct a cardiovascular and cerebrovascular network model with ventricular, arterial and brain regions as nodes and blood flow and neural signaling paths as edges. The structural analysis submodule reveals key nodes and functional clusters in the network through centrality analysis, cluster coefficients and modularity analysis. Finally, in the risk identification sub-module, the dynamic network stability assessment method analyzes the adaptation and stability of the heart brain network under different physiological states, and reveals potential heart brain health risks. The analysis results provide deep knowledge about the heart and brain health condition for the patient and scientific basis for developing personalized rehabilitation strategies.

Referring to fig. 9, the rehabilitation process prediction module includes a process modeling sub-module, a prediction analysis sub-module, and a trend prediction sub-module;

The process modeling module analyzes the historical trend of the health data by adopting a linear regression model based on the health state evaluation and the heart brain network analysis result, processes the heart brain network characteristics by combining with a logistic regression model, further captures key change factors in the rehabilitation process, establishes the integral description of the rehabilitation process and generates a rehabilitation process model;

The trend prediction sub-module evaluates the prediction effect of the model by applying a cross-validation method based on the rehabilitation trend prediction result, optimizes model parameters by using a grid search technology, and generates a rehabilitation process prediction result.

In the process modeling module, a rehabilitation process model is established by adopting a linear regression model and a logistic regression model based on the health state evaluation and the heart brain network analysis result. Linear regression models are used in this application to analyze historical trends in health data, for example, linear regression is used to analyze trends in heart rate, blood pressure, etc. indicators over time. Such analysis may reveal long-term variation laws of the health indicator, such as whether there is a trend to continuously rise or fall. At the same time, logistic regression models are used to deal with characteristics of the heart brain network, especially when those characteristics are related to a particular health state (e.g., the presence or absence of cardiovascular disease). The logistic regression can handle classification problems, and the model can predict the occurrence probability of a certain health condition through analysis of the characteristics of the heart and brain network. These models, in combination, can capture key variables in the rehabilitation process, such as which network characteristics and health indicators are most correlated with rehabilitation progress. The generated rehabilitation process model provides the whole description of the rehabilitation process and provides scientific basis for making and adjusting the rehabilitation plan.

In the predictive analysis sub-module, based on a rehabilitation progress model, an autoregressive integral moving average (ARIMA) model is used for carrying out time series analysis to predict future health state changes of the patient. The ARIMA model is a powerful time series prediction tool and integrates three methods of autoregressive, differential and moving average. In this application, the historical health data is first differentially processed to make the data smoother. The model then builds a predictive model using the autoregressive and moving average portions of the data. The autoregressive portion analyzes the relationships between the historical data points, while the moving average portion analyzes the random fluctuations of the data. By the method, the ARIMA model can predict the trend of health indexes such as heart rate and blood pressure of the patient in a future period of time and identify the occurrence of health risks and rehabilitation trends. The generated rehabilitation trend prediction result is very useful for preventing predicted health problems in advance and optimizing a rehabilitation plan.

In the trend prediction sub-module, based on the rehabilitation trend prediction result, a cross-validation method is applied to evaluate the prediction effect of the model, and the grid search technology is utilized to optimize the model parameters. Cross-validation is a statistical method by which the stability and accuracy of a model is assessed by dividing the data into subsets and repeating the training and validation process on those subsets. The method can ensure that the model can obtain good prediction effects on different data sets, and avoid the problem of over fitting. Grid searching is a parameter optimization technique that finds the optimal model parameter combination by systematically searching within a given range of parameters. For example, in the ARIMA model, a grid search may be used to determine the best auto-regressive terms, the number of differences, and the moving average term. By the aid of the technology, prediction accuracy of the model can be further improved, the generated rehabilitation process prediction result is more accurate and reliable, and better rehabilitation guidance is provided for patients.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, a patient is assumed to have a heart history and hypertension problems. Historical health data shows fluctuations in heart rate and blood pressure over time. In the process modeling sub-module, the change trend of heart rate and blood pressure and the relation between heart brain network characteristics and heart attacks are analyzed through linear and logistic regression models. In the predictive analysis sub-module, the ARIMA model is used to predict heart rate and blood pressure trends for patients for the next several months. Finally, in the trend prediction sub-module, the accuracy and reliability of the prediction result are ensured through cross verification and grid search optimization models. These analysis results help medical teams to develop personalized rehabilitation programs for patients aimed at reducing the risk of heart attacks and controlling blood pressure.

Referring to fig. 10, the rehabilitation tracking module includes an effect monitoring sub-module, a progress evaluation sub-module, and an adjustment suggestion sub-module;

the effect monitoring submodule is used for monitoring physiological parameters of a patient in real time, including heart rate and muscle activity, by adopting heart rate variability analysis and an electromyography signal processing algorithm based on a rehabilitation progress prediction result, evaluating the instant change of a rehabilitation effect and generating rehabilitation state real-time monitoring data;

the progress evaluation sub-module is used for tracking the rehabilitation progress by adopting time sequence analysis based on the rehabilitation state real-time monitoring data, evaluating the rehabilitation speed and efficiency by combining variance analysis and regression analysis, performing multi-dimensional quantitative analysis of the rehabilitation process, and generating comprehensive rehabilitation progress evaluation;

The adjustment suggestion sub-module analyzes the trend of rehabilitation data by adopting a machine learning prediction model based on comprehensive evaluation of the rehabilitation progress, adjusts the rehabilitation scheme according to the current state of the patient by combining a path planning algorithm, provides personalized rehabilitation suggestions and generates rehabilitation effect tracking.

In the effect monitoring submodule, based on the rehabilitation progress prediction result, the physiological parameters of the patient are monitored in real time by adopting heart rate variability analysis and an electromyography signal processing algorithm. These parameters include heart rate and muscle activity, etc., which are critical to assessing rehabilitation effects. Heart rate variability analysis is a measure of heart rhythm variation and reveals the activity of the autonomic nervous system and heart health by analyzing time interval variations in the heart rate sequence. Here, a high and frequent analysis of heart rate variability provides a key indicator for assessing rehabilitation effects. Electromyographic signal processing algorithms, on the other hand, are used to monitor and analyze muscle activity. This involves collecting the electrical signals produced by the muscles and evaluating the response and condition of the muscles by filtering, amplifying and decoding these signals. The real-time monitoring data is recorded in the form of a time series, including the time of each beat of the heart rate and the amplitude of the electromyographic signal. The generated real-time monitoring data of the rehabilitation state provides timely feedback for medical professionals, so that the medical professionals can evaluate the effect of the rehabilitation plan in real time and adjust the rehabilitation plan when necessary.

In the progress assessment sub-module, based on the rehabilitation status real-time monitoring data, a time series analysis is used to track rehabilitation progress, and the steps involve the application of statistical methods such as analysis of variance and regression analysis to assess the speed and efficiency of rehabilitation. Analysis of variance is used to compare physiological parameter changes at different time points or different phases of rehabilitation to determine significant differences in the rehabilitation process. Regression analysis is then used to identify the relationship between physiological parameter changes and rehabilitation activity, such as the correlation of heart rate changes with a particular rehabilitation regimen. These analytical methods provide a multi-dimensional quantitative assessment of the rehabilitation process, including speed, extent of improvement and persistence of rehabilitation. The generated comprehensive assessment result of the rehabilitation progress provides important data support for medical teams, helps the medical teams to better understand the effectiveness of the rehabilitation progress, and adjusts the rehabilitation plan according to the specific situation of patients.

In the adjustment suggestion sub-module, based on comprehensive assessment of rehabilitation progress, a machine learning prediction model and a path planning algorithm are adopted to analyze rehabilitation data trend, and personalized rehabilitation suggestions are provided. The machine learning prediction model predicts a future rehabilitation path by analyzing the historical trend and the current state of rehabilitation data, and identifies a predicted improvement point. For example, models find that a particular type of rehabilitation training is associated with faster health improvement. The path planning algorithm is used for designing a personalized rehabilitation scheme, and taking the current health state, preference and target of the patient into consideration, so as to provide an optimal rehabilitation path for the patient. The generated rehabilitation effect tracking result not only provides a customized rehabilitation plan for the patient, but also ensures the effectiveness and adaptability of the rehabilitation plan through continuous tracking and adjustment.

In the cardiovascular and cerebrovascular rehabilitation intelligent auxiliary system, it is assumed that the patient is undergoing rehabilitation therapy to improve cardiovascular function. In the effect monitoring submodule, the system monitors the heart rate and muscle activity of the patient in real time, and records the physiological response of the patient in real time through heart rate variability analysis and electromyography signal processing. In the progress assessment sub-module, time series analysis reveals the trend of the patient's heart rate and muscle activity during rehabilitation, analysis of variance and regression analysis help the medical team assess the effectiveness of the rehabilitation program. Finally, in the adjustment suggestion sub-module, based on the prediction and path planning algorithm of the machine learning model, the system provides an adjusted personalized rehabilitation scheme for the patient so as to achieve a better rehabilitation effect. These analyses and adjustments help the patient achieve better health improvement during rehabilitation while ensuring personalization and fitness of the rehabilitation program.

The present invention is not limited to the above embodiments, and any equivalent embodiments which can be changed or modified by the technical disclosure described above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above embodiments according to the technical matter of the present invention will still fall within the scope of the technical disclosure.

Claims

1. An intelligent auxiliary system for cardiovascular and cerebrovascular rehabilitation is characterized in that: the system comprises a physiological data monitoring module, a health state analysis module, a behavior pattern recognition module, a rehabilitation scene simulation module, a real-time data processing module, a heart brain network analysis module, a rehabilitation process prediction module and a rehabilitation tracking module;

The rehabilitation tracking module is used for tracking the rehabilitation effect by utilizing a progress monitoring technology based on the rehabilitation progress prediction result, then quantitatively analyzing by applying an effect evaluation method, optimizing the rehabilitation path by adjusting a scheme, and generating a rehabilitation effect tracking;

The health state analysis module comprises a data analysis sub-module, a state evaluation sub-module and a trend analysis sub-module;

the trend analysis submodule carries out trend analysis on the data by adopting linear regression based on trend identification and change point analysis results, determines the key development direction of the health state, and simultaneously smoothes the data sequence by applying a sliding average technology to generate health state assessment;

The behavior pattern recognition module comprises an activity data analysis sub-module, a pattern mining sub-module and an influence factor recognition sub-module;

the influence factor identification submodule classifies the behavior patterns by adopting a K-means clustering algorithm based on the behavior association pattern result, reveals the structure and the type of the behavior patterns by dividing the behavior patterns into differentiated categories, evaluates the statistical difference of the differentiated behavior patterns on health by using variance analysis, thereby determining the influence factors of the behavior patterns on health and generating a behavior pattern influence factor analysis result;

The rehabilitation scene simulation module comprises a strategy simulation sub-module, an effect evaluation sub-module and a simulation feedback sub-module;

The simulation feedback sub-module performs effect feedback optimization based on a strategy applicability evaluation result, adjusts rehabilitation strategy selection, and comprises the steps of comparing differentiated rehabilitation strategies and selecting an optimal strategy, and simultaneously generating rehabilitation strategy simulation by referring to the condition and preference of a patient;

the real-time data processing module comprises an edge computing sub-module, a data flow analysis sub-module and a real-time feedback sub-module;

The real-time feedback submodule analyzes and identifies abnormal modes in real-time data, including abnormal heart rate fluctuation or blood pressure change, based on real-time data flow analysis results by adopting an abnormal detection algorithm based on statistics and machine learning, and timely identifies potential health risks to generate instant health monitoring feedback;

the heart brain network analysis module comprises a network construction sub-module, a structure analysis sub-module and a risk identification sub-module;

The risk identification submodule adopts a dynamic network stability evaluation method based on the network key node analysis result, tracks dynamic changes among nodes by using time sequence analysis, evaluates the stability of the whole network by using a dynamic graph model and generates a heart brain network analysis result;

the rehabilitation process prediction module comprises a process building module, a prediction analysis sub-module and a trend prediction sub-module;

The trend prediction submodule evaluates the prediction effect of the model by applying a cross verification method based on the rehabilitation trend prediction result, optimizes model parameters by using a grid search technology, and generates a rehabilitation process prediction result;

The rehabilitation tracking module comprises an effect monitoring sub-module, a progress evaluation sub-module and an adjustment suggestion sub-module;

2. The intelligent cardiovascular and cerebrovascular rehabilitation assistance system according to claim 1, wherein: the standardized physiological data comprises heart rate waveforms, blood pressure values and respiratory rate, the health state evaluation comprises abnormal value alarms, health indexes and trend prediction, the behavior pattern influence factor analysis results comprise activity habit classification, daily behavior association and health influence score, the rehabilitation strategy simulation comprises strategy effect score, strategy adaptability evaluation and potential risk recognition, the instant health monitoring feedback comprises real-time health alarms, instant parameter changes and response suggestions, the heart brain network analysis results comprise network connection diagrams, key node analysis and network stability evaluation, the rehabilitation process prediction results comprise future health state trends, potential risk evaluation and intervention opportunity judgment, and the rehabilitation effect tracking comprises rehabilitation progress tracking, effect quantitative analysis and strategy adjustment suggestion.

3. The intelligent cardiovascular and cerebrovascular rehabilitation assistance system according to claim 1, wherein: the physiological data monitoring module comprises a data acquisition sub-module, a signal processing sub-module and a data standardization sub-module;