Risk assessment screening system for health management of stroke patient

Abstract

The invention relates to the technical field of medical care, in particular to a risk assessment screening system for health management of a stroke patient, which comprises a data acquisition module, a data preprocessing module, a stroke early-stage micro blood flow change detection module, a stroke risk assessment module and a personalized intervention recommendation module, the data acquisition module is used for acquiring multi-dimensional data of a patient; the data preprocessing module is used for preprocessing the collected multi-dimensional data; the cerebral apoplexy early stage micro blood flow change detection module predicts the cerebral apoplexy early stage risk of the patient; the cerebral apoplexy risk assessment module generates a cerebral apoplexy risk level; and the personalized intervention recommendation module provides personalized health intervention suggestions for the patient. According to the method, potential high-risk patients can be identified, missing of key crowds is avoided, unnecessary intervention can be reduced for low-risk patients, and hierarchical management of risks is effectively realized.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care, and in particular to a risk assessment and screening system for health management of stroke patients. Background Art

[0002] Stroke is one of the leading causes of death and long-term disability worldwide. It is characterized by high morbidity, high mortality and high disability rate, which seriously threatens human health and quality of life. Stroke is hidden and sudden, and early clinical symptoms are often not obvious. Traditional health management and screening methods are difficult to effectively predict and intervene before the onset of the disease. Therefore, how to achieve accurate assessment and early warning of early stroke risks and reduce the risk of stroke through effective health management methods has become an important issue that needs to be urgently addressed in the field of medical health management.

[0003] At present, stroke risk assessment mainly relies on a single assessment of static health data (such as blood pressure, blood sugar, heart rate, etc.), and lacks real-time monitoring and analysis of patients' dynamic physiological states (such as changes in cerebral blood flow). In addition, existing risk screening methods are deficient in data integrity, processing efficiency, and prediction accuracy, making it difficult to effectively identify high-risk patients. Traditional health intervention programs often lack personalized considerations and fail to provide targeted intervention measures based on the patient's actual risk level and health status, resulting in limited early prevention and control effects and difficulty in achieving precise and dynamic health management.

[0004] The present invention aims to overcome the deficiencies of the prior art and provide a risk assessment and screening system for the health management of stroke patients, which can effectively reduce the risk of stroke and improve the health management level and quality of life of patients. Summary of the invention

[0005] The invention provides a risk assessment and screening system for health management of stroke patients.

[0006] A risk assessment and screening system for health management of stroke patients, comprising a data acquisition module, a data preprocessing module, an early stroke microblood flow change detection module, a stroke risk assessment module and a personalized intervention recommendation module, wherein;

[0007] The data collection module collects multi-dimensional data of the patient, including health data, environmental information and social behavior data;

[0008] The data preprocessing module preprocesses the collected multi-dimensional data, including filling missing data and removing noise;

[0009] The early stroke micro-blood flow change detection module collects the patient's cerebral blood flow data in real time, and analyzes the cerebral blood flow data through a convolutional neural network (CNN) model to identify abnormal fluctuations in cerebral blood flow and predict the patient's early stroke risk, specifically including:

[0010] Cerebral blood flow data collection: Real-time monitoring of the patient's cerebral blood flow to obtain physiological parameters of cerebral blood flow, including cerebral blood flow velocity, blood flow, and blood oxygen saturation;

[0011] Cerebral blood flow data analysis: The collected cerebral blood flow data is analyzed through the convolutional neural network (CNN) model to identify abnormal fluctuations in blood flow;

[0012] Early risk prediction of stroke: Based on the identified abnormal fluctuations in blood flow and combined with the patient's baseline health data, the patient's early risk of stroke is predicted;

[0013] The stroke risk assessment module comprehensively assesses the patient's stroke risk based on the preprocessed multi-dimensional data and the prediction results of the patient's early stroke risk, and generates a stroke risk level, including low risk, medium risk and high risk;

[0014] The personalized intervention recommendation module provides patients with personalized health intervention suggestions based on the generated stroke risk level, including diet adjustment, exercise plan, drug use, and psychological intervention.

[0015] Optionally, the data acquisition module includes:

[0016] Health data collection: Medical monitoring equipment (heart rate monitor, blood pressure monitor, blood oxygen saturation meter, blood glucose meter, thermometer) is used to collect patients' health data in real time, including heart rate, blood pressure, blood oxygen, blood glucose, weight and body temperature;

[0017] Environmental information collection: The air quality index (AQI), PM2.5, temperature, humidity and air pressure of the patient's environment are monitored in real time through air quality monitors and climate condition sensors;

[0018] Social behavior data collection: Use smart devices (smart watches, mobile phones, or smart home devices) to record the patient’s daily social activity frequency, interaction time with family members, and supportive family environment.

[0019] Optionally, the data preprocessing module includes:

[0020] Missing data filling: For missing values ​​that occur during multi-dimensional data collection, linear interpolation is used to fill them;

[0021] Denoising: Wavelet transform is used to reduce signal noise of multi-dimensional data.

[0022] Optionally, the cerebral blood flow data acquisition includes:

[0023] Real-time monitoring of cerebral blood flow velocity: Use brain ultrasound sensors (transcranial Doppler ultrasound, TCD) to monitor cerebral blood flow in real time, and calculate the cerebral blood flow velocity v through the ultrasonic reflection signal;

[0024] Calculation and monitoring of blood flow: Obtain blood vessel cross-sectional area through magnetic resonance angiography, and automatically calculate brain blood flow Q based on blood flow velocity;

[0025] Real-time detection of blood oxygen saturation: The near-infrared spectroscopy sensor (NIRS) is used to detect the blood oxygen level in the brain, and the local blood oxygen saturation SpO2 in the brain is calculated by using the different absorption rates of hemoglobin to light of different wavelengths.

[0026] Optionally, the convolutional neural network (CNN) model includes:

[0027] Input format optimization: Continuous wavelet transform is used to perform wavelet transform on the collected cerebral blood flow data, and the time and frequency characteristics of cerebral blood flow data at different scales are extracted as input data for the convolutional neural network (CNN) model;

[0028] Dynamic Adaptation of Convolutional Kernels: Multi-scale convolutional kernels are used to extract local details and global features, and an attention mechanism is introduced to dynamically assign weights to convolutional kernels of different scales. At the same time, for the same input data, the feature maps output by convolutional kernels of all scales are weighted and fused;

[0029] Activation function optimization: Use smooth ReLU activation function to capture tiny features;

[0030] Output layer: The temporal attention mechanism is introduced to predict abnormal microblood flow fluctuations and output the probability distribution of abnormal fluctuations P(y=c|x).

[0031] Optionally, the early risk prediction of stroke includes:

[0032] Data fusion: weighted combination of abnormal probability output from cerebral blood flow data analysis and baseline health data to generate a comprehensive feature vector F;

[0033] Early risk prediction: Logistic regression is used to calculate the comprehensive feature vector F to predict the patient's early stroke risk probability P (y = 1|F);

[0034] Early risk judgment: According to the predicted early risk probability P(y=1|F), the patient's early stroke risk is divided. When P(y=1|F)≤0.5, it means that the patient currently has no early stroke risk. When P(y=1|F)>0.5, it means that the patient currently has early stroke risk.

[0035] Optionally, the stroke risk assessment module includes:

[0036] Multidimensional data analysis: Comprehensively process the pre-processed multidimensional data of patients to generate comprehensive data for evaluation risk ;

[0037] Risk classification: Based on the comprehensive data generated for assessment risk , combined with the results of early stroke risk prediction (no early stroke risk, early stroke risk), patients are divided into risk levels, including low risk, medium risk, and high risk.

[0038] Optionally, the multi-dimensional data analysis includes:

[0039] Data normalization: normalize the pre-processed multi-dimensional data of patients;

[0040] Data comprehensive processing: weighted fusion of normalized multi-dimensional patient data to generate comprehensive data F for evaluation risk .

[0041] Optionally, the risk level classification includes:

[0042] Low risk: When the comprehensive data F risk ≤ the lower limit of the risk threshold T1, and the result of early stroke risk prediction is that the patient currently has no early stroke risk, the patient is assessed as low risk;

[0043] Medium risk: When the lower limit of risk threshold T1 < comprehensive data F risk ≤ the upper limit of the risk threshold T2, or the result of the early stroke risk prediction is that the patient currently has no early stroke risk but the lower limit of the risk threshold T1 < the comprehensive data F risk When , the patient is assessed as medium risk;

[0044] High risk: When the comprehensive data F risk When the risk threshold value is > upper limit T2, or the result of early stroke risk prediction indicates that the patient currently has early stroke risk, the patient is assessed as high risk.

[0045] Optionally, the personalized intervention recommendation module includes:

[0046] Diet adjustment: For patients at medium and high risk, we recommend a diet plan based on their health data, including a low-salt, low-fat diet and controlled sugar intake;

[0047] Exercise plan: Generate an exercise plan based on the patient's risk level and physical condition. For low-risk patients, aerobic exercise is recommended, including walking and jogging. For medium-risk patients, a light to moderate exercise plan is provided, including yoga and swimming. For high-risk patients, rehabilitation training is recommended and supervised.

[0048] Medication use: For patients at medium and high risk, guidance on the use of antihypertensive and antiplatelet drugs is provided in combination with the doctor's advice;

[0049] Psychological intervention: Provide psychological assessment for high-risk patients and recommend relaxation training and psychological counseling.

[0050] Beneficial effects of the present invention:

[0051] The present invention, through a multi-dimensional data acquisition module, integrates the patient's health data, environmental information and social behavior data to form a comprehensive individual health portrait, and combines the missing data filling and denoising processing of the data preprocessing module to ensure the integrity, accuracy and consistency of the data. This high-quality data input provides a solid data foundation for abnormal cerebral blood flow fluctuation detection and risk prediction, enabling the system to capture subtle changes in the patient's health status and enhance the reliability and stability of risk screening.

[0052] The present invention effectively combines static baseline health data with dynamic cerebral blood flow data through an early stroke microblood flow change detection module and a stroke risk assessment module, identifies abnormal fluctuations in cerebral blood flow and predicts the early risk of patients, and simultaneously performs weighted fusion and multi-dimensional analysis on the comprehensive data of patients to accurately divide the risk level. This multi-dimensional data analysis and risk level division mechanism can not only identify potential high-risk patients and avoid missing key populations, but also reduce unnecessary interventions for low-risk patients, effectively realize hierarchical management of risks, and provide a scientific basis for clinical early warning and intervention decisions.

[0053] The present invention provides targeted health intervention suggestions, including diet adjustment, exercise plan, drug use and psychological intervention, through a personalized intervention recommendation module combined with the patient's risk level, to form a comprehensive health management plan. It can classify and accurately intervene according to the patient's actual risk status, ensure that low-risk patients maintain a healthy state, medium-risk patients are effectively improved, and high-risk patients receive enhanced management and medical supervision. Through multi-dimensional and dynamic risk assessment and intervention measures, accurate prediction of stroke risk, real-time warning and personalized health management are achieved, which significantly reduces the incidence of stroke and improves the patient's quality of life and clinical prevention and control effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 A schematic diagram of system function modules according to an embodiment of the present invention;

[0056] Figure 2 Schematic diagram of a stroke early microblood flow change detection module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0058] like Figure 1-Figure 2 As shown, a risk assessment and screening system for health management of stroke patients includes a data acquisition module, a data preprocessing module, an early stroke microblood flow change detection module, a stroke risk assessment module and a personalized intervention recommendation module, wherein;

[0059] The data collection module collects multi-dimensional data of patients, including health data, environmental information, and social behavior data;

[0060] The data preprocessing module preprocesses the collected multi-dimensional data, including filling missing data and removing noise;

[0061] The early stroke micro-blood flow change detection module collects the patient's cerebral blood flow data in real time, analyzes the cerebral blood flow data through the convolutional neural network (CNN) model, identifies abnormal fluctuations in cerebral blood flow, and predicts the patient's early stroke risk, including:

[0062] Cerebral blood flow data collection: Real-time monitoring of the patient's cerebral blood flow to obtain physiological parameters of cerebral blood flow, including cerebral blood flow velocity, blood flow, and blood oxygen saturation;

[0063] Cerebral blood flow data analysis: The collected cerebral blood flow data is analyzed through the convolutional neural network (CNN) model to identify abnormal fluctuations in blood flow;

[0064] Early risk prediction of stroke: Based on the identified abnormal fluctuations in blood flow and combined with the patient's baseline health data, the patient's early risk of stroke is predicted;

[0065] The stroke risk assessment module conducts a comprehensive assessment of the patient's stroke risk based on the pre-processed multi-dimensional data and the prediction results of the patient's early stroke risk, and generates a stroke risk level, including low risk, medium risk and high risk;

[0066] The personalized intervention recommendation module provides patients with personalized health intervention suggestions based on the generated stroke risk level, including diet adjustment, exercise plan, medication use, and psychological intervention;

[0067] Through the above content, it is possible to comprehensively and accurately assess the early risk of stroke and provide customized health intervention plans. It combines the patient's physiological parameters, environmental information and social behavior data, so as to achieve accurate prediction and early warning of stroke risk, effectively identify potential health threats, and provide dynamic and real-time risk assessment and intervention recommendations, thereby significantly improving the accuracy and timeliness of stroke prevention and control, improving the patient's health management level, and reducing the risk of stroke.

[0068] The data acquisition module includes:

[0069] Health data collection: Medical monitoring equipment (heart rate monitor, blood pressure monitor, blood oxygen saturation meter, blood glucose meter, thermometer) is used to collect patients' health data in real time, including heart rate, blood pressure, blood oxygen, blood glucose, weight and body temperature;

[0070] Environmental information collection: The air quality index (AQI), PM2.5, temperature, humidity and air pressure of the patient's environment are monitored in real time through air quality monitors and climate condition sensors;

[0071] Social behavior data collection: Record the patient's daily social activity frequency, interaction time with family members, and supportive family environment through smart devices (smart watches, mobile phones, or smart home devices);

[0072] Through the above content, the patient's health data, environmental information and social behavior data are comprehensively collected, providing a multi-dimensional, real-time health monitoring platform that can comprehensively reflect the patient's physical condition, living environment and social interaction, and provide rich basic data support for subsequent risk assessment and personalized intervention.

[0073] The data preprocessing module includes:

[0074] Missing data filling: For missing values ​​that occur during multi-dimensional data collection, linear interpolation is used to fill them, expressed as:

[0075]

[0076] Where x(t) is the missing value, t-1 and t+1 are the data points before and after the missing value, x(t-1) is the data value at time t-1, and x(t+1) is the data value at time t+1;

[0077] Denoising: In order to remove the noise in the sensor data, wavelet transform is used to perform signal denoising on multi-dimensional data, which can be expressed as:

[0078]

[0079] Among them, ψ(t) is the mother wavelet, a is the scale parameter, and b is the translation parameter;

[0080] Through the above content, the integrity, accuracy and consistency of the collected data are ensured, which provides high-quality data input for subsequent stroke risk assessment and analysis. It can filter out irrelevant noise, retain key information, improve the system's processing efficiency and accuracy of multi-dimensional data, and ensure the reliability and stability of stroke risk assessment and early warning results.

[0081] Cerebral blood flow data collection includes:

[0082] Real-time monitoring of cerebral blood flow velocity: The cerebral blood flow is monitored in real time using a brain ultrasound sensor (transcranial Doppler ultrasound, TCD), and the cerebral blood flow velocity v is calculated by the ultrasonic reflection signal, which is expressed as:

[0083]

[0084] Where v is the cerebral blood flow velocity, f d is the frequency shift, c is the propagation speed of sound waves in tissues, f0 is the ultrasonic emission frequency, and θ is the angle between the ultrasonic wave and the blood flow direction;

[0085] Calculation and monitoring of blood flow: The cross-sectional area of ​​blood vessels is obtained through magnetic resonance angiography, and the brain blood flow Q is automatically calculated based on the blood flow velocity, expressed as:

[0086] Q = v·A;

[0087] Among them, Q is the blood flow rate, v is the blood flow velocity, and A is the cross-sectional area of ​​the blood vessel;

[0088] Real-time detection of blood oxygen saturation: The near-infrared spectroscopy sensor (NIRS) is used to detect the blood oxygen level in the brain. The different absorption rates of hemoglobin to light of different wavelengths are used to calculate the local blood oxygen saturation SpO2 in the brain, which is expressed as:

[0089]

[0090] Among them, SpO2 is blood oxygen saturation, HbO2 is oxygenated hemoglobin concentration, and Hb is deoxygenated hemoglobin concentration;

[0091] Through the above content, real-time monitoring of patients' cerebral blood flow is achieved, and key physiological parameters such as cerebral blood flow velocity, blood flow and blood oxygen saturation are accurately obtained. It can not only dynamically capture the microscopic changes of cerebral blood flow, but also has the advantages of high timeliness and non-invasiveness, ensuring the continuity and accuracy of data, and providing reliable data support for the accurate assessment and early warning of early stroke risks, effectively improving the efficiency of identifying potential stroke risks and the clinical application value.

[0092] The Convolutional Neural Network (CNN) model includes:

[0093] Input format optimization: Cerebral blood flow data has time series characteristics and contains noise. It is necessary to reconstruct the data and convert it into a format suitable for CNN input (time-frequency two-dimensional feature map). The collected cerebral blood flow data is transformed by continuous wavelet transform to extract the time and frequency characteristics of cerebral blood flow data at different scales as the input data of the convolutional neural network (CNN) model, which is expressed as:

[0094]

[0095] Among them, W(a,b) is the wavelet transform coefficient, ψ * is the complex conjugate of the mother wavelet function, a and b represent the scale and translation parameters, respectively, x(t) is the cerebral blood flow data, and t is the time variable;

[0096] Dynamic adaptation of convolution kernels: Multi-scale convolution kernels are used to extract local details and global features, and an attention mechanism is introduced to dynamically assign weights to convolution kernels of different scales. At the same time, for the same input data, the feature maps output by convolution kernels of all scales are weighted and fused, which can be expressed as:

[0097]

[0098] Among them, x(i,j) is the input data (time-frequency diagram), k(i,j) is the convolution kernel, h and w are the height and width of the convolution kernel, K m,n is the convolution result;

[0099] k adaptive (i,j)=α·k small (i,j)+β·k large (i,j);

[0100] Among them, k small (i,j) and k large (i, j) are small and large convolution kernels, α and β are weight coefficients (α+β=1), k adaptive (i, j) is the convolution kernel k of different scales at the position (i, j) of the input data.small (i,j) and k large (i, j) The final convolution kernel output after weighted fusion;

[0101] Activation function optimization: Since the abnormal fluctuation amplitude in the cerebral blood flow signal is small, a smooth ReLU activation function is used to capture small features, which is expressed as:

[0102]

[0103] Among them, x(t) represents the characteristic value of the input cerebral blood flow data at time t, δ is the smoothing parameter, and f(x(t)) is the output result of the smoothed ReLU;

[0104] Output layer: The temporal attention mechanism is introduced to predict abnormal microblood flow fluctuations and output the probability distribution of abnormal fluctuations P(y=c|x), which is expressed as:

[0105]

[0106] in, is the attention weight, x i is the feature of each time step, A t is the weighted time feature, T is the total number of time steps;

[0107]

[0108] Among them, W c and b c are the weight and bias of category c, respectively, C is the number of categories (normal, abnormal fluctuations), P(y=c|x) is the probability that the input data x is classified as category c, and W j and b j are the weight and bias of category j respectively;

[0109] Through the above content, it is possible to capture local subtle changes and global features at the same time, and effectively identify the key time points of abnormal fluctuations in cerebral blood flow. The smoothed ReLU activation function enhances the model's sensitivity to small-amplitude signals, and the temporal attention mechanism further focuses on important timing features related to abnormalities, thereby improving the detection accuracy and robustness of abnormal fluctuations. The model outputs the probability distribution of abnormal fluctuations through the Softmax classifier, providing more accurate and reliable data support for early risk prediction of stroke, and significantly improving the system's ability to identify potential risks of stroke and its clinical application value.

[0110] Early risk prediction for stroke includes:

[0111] Data fusion: The abnormal probability output from the cerebral blood flow data analysis and the baseline health data are weightedly combined to generate a comprehensive feature vector F, which is expressed as:

[0112] F=W1·P a +W2·f h ealth +b;

[0113] Among them, P a is the output abnormal probability value, W1 and W2 represent the importance weights of abnormal probability and baseline health data respectively, b is the bias term, and f h ealth for baseline health data;

[0114] Early risk prediction: Logistic regression is used to calculate the comprehensive feature vector F to predict the patient's early stroke risk probability P (y = 1 | F), expressed as:

[0115]

[0116] Among them, P(y=1|F) represents the early risk probability of stroke in patients, W3 is the weight vector of the logistic regression model, and b′ is the bias term;

[0117] Early risk judgment: According to the predicted early risk probability P(y=1|F), the patient's early stroke risk is divided. When P(y=1|F)≤0.5, it means that the patient currently has no early stroke risk. When P(y=1|F)>0.5, it means that the patient currently has early stroke risk.

[0118] Baseline health dataf h ealth It is a feature vector composed of physiological indicators within the normal range, including:

[0119] (1) Heart rate (f hr ) indicates the number of heart beats per minute (unit: beats / minute), the normal range is 60-100 beats / minute;

[0120] (2) Blood pressure (f bp ) includes systolic and diastolic blood pressure, which reflects the pressure on the arteries. The normal range is systolic pressure 90-120 mmHg and diastolic pressure 60-80 mmHg;

[0121] (3) Blood oxygen saturation (f spO2 ) indicates the proportion of oxygenated hemoglobin in the blood, reflecting the oxygen supply capacity of the blood, with a normal range of 95%-100%;

[0122] (4) Blood sugar (f glucose ) reflects the concentration of glucose in the blood and is an important indicator for evaluating metabolic status. The normal range is fasting blood glucose 3.9-6.1mmol / L;

[0123] (5)Weight (f weig h t) represents the patient's weight (unit: kg), which is used as a reference for assessing obesity or emaciation;

[0124] (6)Body temperature (f temp ), which indicates the patient's body temperature (unit: °C), reflecting the basal metabolic level, with a normal range of 36.5-37.5 °C;

[0125] Through the above content, a comprehensive analysis of static physiological characteristics and dynamic abnormal cerebral blood flow fluctuations is conducted to provide patients with an accurate risk screening mechanism. By weighting the characteristics through a logistic regression model and generating risk probabilities, patients can be quickly and efficiently divided into those with early risk or without early risk, thereby achieving preliminary screening of potential stroke risks. This not only improves the accuracy and reliability of stroke risk prediction, but also significantly improves the efficiency of early screening.

[0126] The stroke risk assessment module includes:

[0127] Multidimensional data analysis: Comprehensively process the pre-processed multidimensional data of patients to generate comprehensive data for evaluation risk ;

[0128] Risk classification: Based on the comprehensive data generated for assessment risk , combined with the results of early stroke risk prediction (no early stroke risk, early stroke risk), the patients are divided into risk levels, including low risk, medium risk, and high risk;

[0129] Through the above content, the multi-dimensional data of the patients after pretreatment are comprehensively processed to generate quantitative comprehensive data, and combined with the early risk prediction results of stroke (such as no early risk, early risk), the patients are accurately divided into risk levels (low risk, medium risk, high risk). By using the patient's health data and early risk information, static health characteristics and dynamic cerebral blood flow change characteristics are effectively integrated to achieve a comprehensive assessment of stroke risk, accurately identify high-risk patients, provide graded health risk assessment, and provide a scientific basis for early intervention and personalized health management, which will help reduce the incidence of stroke and the medical burden.

[0130] Multi-dimensional data analysis includes:

[0131] Data normalization: The pre-processed multi-dimensional data of patients are normalized to eliminate the influence of different feature dimensions on the comprehensive evaluation results and ensure that the data is analyzed at a unified scale, which is expressed as:

[0132]

[0133] Among them, f i normis the normalized multi-dimensional data, f i is the original multi-dimensional data, f i min and f i max They are multi-dimensional data f i The minimum and maximum values ​​of

[0134] Data comprehensive processing: weighted fusion of normalized multi-dimensional patient data to generate comprehensive data F for evaluation risk , expressed as:

[0135]

[0136] Among them, w i is the weight of each multi-dimensional data, n is the number of multi-dimensional data;

[0137] Through the above content, the influence of dimensional differences between different features is eliminated, so that the data can be analyzed at a unified scale. At the same time, the comprehensive data F is generated through the weighted fusion method. risk , comprehensively reflecting the patient's overall health status.

[0138] The risk levels include:

[0139] Low risk: When the comprehensive data F risk ≤ the lower limit of the risk threshold T1, and the result of early stroke risk prediction is that the patient currently has no early stroke risk, the patient is assessed as low risk;

[0140] Medium risk: When the lower limit of risk threshold T1 < comprehensive data F risk ≤ the upper limit of the risk threshold T2, or the result of the early stroke risk prediction is that the patient currently has no early stroke risk but the lower limit of the risk threshold T1 < the comprehensive data F risk When , the patient is assessed as medium risk;

[0141] High risk: When the comprehensive data F risk >The upper limit of the risk threshold T2, or when the result of early stroke risk prediction indicates that the patient currently has an early stroke risk, the patient is assessed as high risk;

[0142] The lower risk threshold T1 and the upper risk threshold T2 are set based on historical data and statistical methods, including:

[0143] Data collection: Collect a large amount of historical data with annotated stroke risk levels, including multi-dimensional data (heart rate, blood pressure, blood sugar, etc.) and early risk probability;

[0144] Calculate the comprehensive data distribution: Calculate the comprehensive data F risk ;

[0145] Determine the quantile of the distribution: Calculate the comprehensive data F risk The distribution of and the corresponding quantiles are taken as thresholds, T1 is the 30% quantile, T2 is the 70% quantile;

[0146] Through the above content, low-risk, medium-risk and high-risk patients can be accurately distinguished, and static health data can be combined with dynamic early risk signals to ensure the comprehensiveness and scientificity of the assessment results. By setting risk thresholds, the boundaries of different risk levels can be effectively captured, which can not only identify patients in good health and avoid excessive intervention, but also accurately screen out medium- and high-risk patients, providing a reliable basis for early intervention. In addition, the division rules have clear logic and comprehensive coverage, which can adapt to the risk assessment needs of different patient groups and help improve the efficiency and accuracy of stroke warning and management.

[0147] The personalized intervention recommendation module includes:

[0148] Diet adjustment: For patients at medium and high risk, we recommend a diet plan based on their health data, including a low-salt, low-fat diet and controlled sugar intake;

[0149] Exercise plan: Generate an exercise plan based on the patient's risk level and physical condition. For low-risk patients, aerobic exercise is recommended, including walking and jogging. For medium-risk patients, a light to moderate exercise plan is provided, including yoga and swimming. For high-risk patients, rehabilitation training is recommended and supervised.

[0150] Medication use: For patients at medium and high risk, guidance on the use of antihypertensive and antiplatelet drugs is provided in combination with the doctor's advice;

[0151] Psychological intervention: Provide psychological assessment for high-risk patients, recommend relaxation training and psychological counseling to reduce patients' psychological stress and anxiety and improve their overall health;

[0152] Through the above content, targeted dietary adjustments, exercise plans, medication use and psychological intervention programs are provided to achieve hierarchical and precise health management. Intervention measures can be tailored according to the patient's actual risk status to ensure that low-risk patients remain healthy, medium-risk patients receive timely health improvement, and high-risk patients receive intensive intervention and professional supervision. This module covers multiple dimensions of diet, exercise, medication and psychology, forming a comprehensive health intervention system, which effectively reduces the risk of stroke, improves the patient's quality of life and health management efficiency, and has significant clinical application value.

[0153] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0154] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A risk assessment and screening system for health management of stroke patients, characterized in that: It includes data acquisition module, data preprocessing module, early stroke microblood flow change detection module, stroke risk assessment module and personalized intervention recommendation module, among which; The data collection module collects multi-dimensional data of the patient, including health data, environmental information and social behavior data; The data preprocessing module preprocesses the collected multi-dimensional data, including filling missing data and removing noise; The early stroke micro-blood flow change detection module collects the patient's cerebral blood flow data in real time, and analyzes the cerebral blood flow data through a convolutional neural network model to identify abnormal fluctuations in cerebral blood flow and predict the patient's early stroke risk, specifically including: Cerebral blood flow data collection: Real-time monitoring of the patient's cerebral blood flow to obtain physiological parameters of cerebral blood flow, including cerebral blood flow velocity, blood flow, and blood oxygen saturation; Cerebral blood flow data analysis: The collected cerebral blood flow data is analyzed through the convolutional neural network model to identify abnormal fluctuations in blood flow; Early risk prediction of stroke: Based on the identified abnormal fluctuations in blood flow and combined with the patient's baseline health data, the patient's early risk of stroke is predicted; The stroke risk assessment module comprehensively assesses the patient's stroke risk based on the preprocessed multi-dimensional data and the prediction results of the patient's early stroke risk, and generates a stroke risk level, including low risk, medium risk and high risk; The personalized intervention recommendation module provides patients with personalized health intervention suggestions based on the generated stroke risk level, including diet adjustment, exercise plan, drug use, and psychological intervention.

2. A risk assessment and screening system for health management of stroke patients according to claim 1, characterized in that: The data acquisition module comprises: Health data collection: Medical monitoring equipment is used to collect patients’ health data in real time, including heart rate, blood pressure, blood oxygen, blood sugar, weight and body temperature; Environmental information collection: real-time monitoring of the patient's environment's air quality index, PM2.5, temperature, humidity and air pressure through air quality monitors and climate condition sensors; Social behavior data collection: Smart devices are used to record the patient’s daily social activity frequency, interaction duration with family members, and supportiveness of the family environment.

3. A risk assessment and screening system for health management of stroke patients according to claim 1, characterized in that: The data preprocessing module comprises: Missing data filling: For missing values ​​that occur during multi-dimensional data collection, linear interpolation is used to fill them; Denoising: Wavelet transform is used to reduce signal noise of multi-dimensional data.

4. A risk assessment and screening system for health management of stroke patients according to claim 1, characterized in that: The cerebral blood flow data acquisition includes: Real-time monitoring of cerebral blood flow velocity: Use brain ultrasound sensors to monitor cerebral blood flow in real time, and calculate the cerebral blood flow velocity v through ultrasonic reflection signals; Calculation and monitoring of blood flow: Obtain blood vessel cross-sectional area through magnetic resonance angiography, and automatically calculate brain blood flow Q based on blood flow velocity; Real-time detection of blood oxygen saturation: The brain's blood oxygen level is detected by a near-infrared spectral sensor, and the local brain blood oxygen saturation SpO2 is calculated by using the different absorption rates of hemoglobin to light of different wavelengths.

5. A risk assessment and screening system for health management of stroke patients according to claim 4, characterized in that: The convolutional neural network model includes: Input format optimization: Continuous wavelet transform is used to perform wavelet transform on the collected cerebral blood flow data, and the time and frequency characteristics of cerebral blood flow data at different scales are extracted as input data for the convolutional neural network model; Dynamic Adaptation of Convolutional Kernels: Multi-scale convolutional kernels are used to extract local details and global features, and an attention mechanism is introduced to dynamically assign weights to convolutional kernels of different scales. At the same time, for the same input data, the feature maps output by convolutional kernels of all scales are weighted and fused; Activation function optimization: Use smooth ReLU activation function to capture tiny features; Output layer: The temporal attention mechanism is introduced to predict abnormal microblood flow fluctuations and output the probability distribution of abnormal fluctuations P(y=c|x).

6. A risk assessment and screening system for health management of stroke patients according to claim 5, characterized in that: The early risk prediction of stroke includes: Data fusion: weighted combination of abnormal probability output from cerebral blood flow data analysis and baseline health data to generate a comprehensive feature vector F; Early risk prediction: Logistic regression is used to calculate the comprehensive feature vector F to predict the patient's early stroke risk probability P (y = 1|F); Early risk judgment: According to the predicted early risk probability P(y=1|F), the patient's early stroke risk is divided. When P(y=1|F)≤0.5, it means that the patient currently has no early stroke risk. When P(y=1|F)>0.5, it means that the patient currently has early stroke risk.

7. A risk assessment and screening system for health management of stroke patients according to claim 6, characterized in that: The stroke risk assessment module includes: Multidimensional data analysis: Comprehensively process the pre-processed multidimensional data of patients to generate comprehensive data for evaluation risk ; Risk classification: Based on the comprehensive data generated for assessment risk , combined with the results of early stroke risk prediction, patients are divided into risk levels, including low risk, medium risk, and high risk.

8. A risk assessment and screening system for health management of stroke patients according to claim 7, characterized in that: The multi-dimensional data analysis includes: Data normalization: normalize the pre-processed multi-dimensional data of patients; Data comprehensive processing: weighted fusion of normalized multi-dimensional patient data to generate comprehensive data F for evaluation risk .

9. A risk assessment and screening system for health management of stroke patients according to claim 8, characterized in that: The risk level classification includes: Low risk: When the comprehensive data F risk ≤ the lower limit of the risk threshold T1, and the result of early stroke risk prediction is that the patient currently has no early stroke risk, the patient is assessed as low risk; Medium risk: When the lower limit of risk threshold T1 < comprehensive data F risk ≤ the upper limit of the risk threshold T2, or the result of the early stroke risk prediction is that the patient currently has no early stroke risk but the lower limit of the risk threshold T1 < the comprehensive data F risk When , the patient is assessed as medium risk; High risk: When the comprehensive data F risk When the risk threshold value is > upper limit T2, or the result of early stroke risk prediction indicates that the patient currently has early stroke risk, the patient is assessed as high risk.

10. A risk assessment and screening system for health management of stroke patients according to claim 1, characterized in that: The personalized intervention recommendation module includes: Diet adjustment: For patients at medium and high risk, we recommend a diet plan based on their health data, including a low-salt, low-fat diet and controlled sugar intake; Exercise plan: Generate an exercise plan based on the patient's risk level and physical condition. For low-risk patients, aerobic exercise is recommended, including walking and jogging. For medium-risk patients, a light to moderate exercise plan is provided, including yoga and swimming. For high-risk patients, rehabilitation training is recommended and supervised. Medication use: For patients at medium and high risk, guidance on the use of antihypertensive and antiplatelet drugs is provided in combination with the doctor's advice; Psychological intervention: Provide psychological assessment for high-risk patients and recommend relaxation training and psychological counseling.

Images