Cardiovascular disease data fusion and multi-dimensional health assessment method

Abstract

The invention relates to the technical field of health monitoring, in particular to a cardiovascular disease data fusion and multi-dimensional health assessment method, which comprises the following steps of: acquiring multi-source data of a patient through a sensor and wearable equipment, preprocessing the acquired data, ensuring data integrity, extracting physiological features by adopting a convolutional neural network, and evaluating the health of the patient. The method comprises the steps of extracting non-physiological data features through principal component analysis, constructing a multi-dimensional feature set, identifying cardiovascular disease related risk factors through a random forest algorithm, distributing weights through a logistic regression algorithm to generate individualized risk scores, analyzing time sequence trends of the risk factors based on a long and short-term memory network model, and generating dynamic risk scores in real time. When the score exceeds a preset threshold value, an early warning mechanism is automatically triggered, and a patient and medical staff are notified through an application program; according to the method, accurate assessment and personalized intervention of cardiovascular health are realized through multi-source data fusion and dynamic analysis, and the method has high efficiency and timeliness.

Description

Technical Field

[0001] The present invention relates to the field of health monitoring technology, and in particular to a cardiovascular disease data fusion and multi-dimensional health assessment method. Background Art

[0002] Currently, cardiovascular disease is one of the major diseases threatening human health worldwide, with its incidence and mortality rates continuing to rise. The causes of cardiovascular disease are complex and diverse, and are often affected by multiple factors such as physiological factors, lifestyle, and environmental factors.

[0003] However, existing technologies still have some technical difficulties in cardiovascular health monitoring and risk assessment. First, the collection of a single data source is not enough to fully cover cardiovascular health risk factors, and it is difficult to integrate multi-source data such as physiological parameters, lifestyle, and environmental impacts, resulting in one-sided and inaccurate assessment results. Secondly, most traditional assessment methods cannot realize dynamic analysis of health data, lack real-time tracking and early warning functions of patient health status, and therefore it is difficult to identify rising risk trends in a timely manner. In addition, existing technologies lack personalized health intervention mechanisms and fail to generate targeted health management recommendations based on dynamic risk scores, resulting in low timeliness and effectiveness of health interventions. Summary of the invention

[0004] The present invention provides a cardiovascular disease data fusion and multi-dimensional health assessment method.

[0005] The cardiovascular disease data fusion and multi-dimensional health assessment method includes the following steps:

[0006] S1, data collection: by deploying a variety of sensors and wearable devices, and combining with the electronic medical record system to obtain multi-source data of patients, the multi-source data includes physiological parameters, historical health records, lifestyle data and environmental parameters;

[0007] S2, data preprocessing: preprocessing the collected multi-source data, including noise removal, missing value filling, and outlier detection;

[0008] S3, feature extraction and data fusion: use convolutional neural network to extract features from physiological parameters, use principal component analysis to extract features from historical health records, lifestyle data and environmental parameters, integrate the features extracted from physiological parameters with the features extracted from historical health records, lifestyle data and environmental parameters to form a multi-dimensional feature set;

[0009] S4, risk factor identification and weighting: The random forest algorithm is used to identify risk factors associated with cardiovascular disease from a multidimensional feature set. At the same time, a logistic regression algorithm is used to assign weights to each risk factor to generate an individualized risk score for preliminary health assessment;

[0010] S5, Predictive modeling and risk assessment: Based on the identified risk factors, a long-short term memory network model is constructed to analyze the time series trends of historical multi-source data and real-time multi-source data, predict the risk of cardiovascular events, and generate dynamic risk scores;

[0011] S6, Personalized advice and intervention: Generate personalized health intervention recommendations based on the dynamic risk score output by the long short-term memory network model, including lifestyle adjustments, drug management, and regular check-up plans. If the dynamic risk score exceeds the preset threshold, the early warning mechanism is automatically triggered to notify patients and medical staff through the application.

[0012] Optionally, the S1 includes:

[0013] S11, Sensor and equipment deployment: Deploy a variety of sensors and wearable devices in the patient's daily living environment to collect physiological parameter data, including electrocardiogram sensors to obtain patients' electrocardiogram signals, blood pressure sensors to monitor patients' blood pressure changes, and heart rate sensors to measure patients' heart rates;

[0014] S12, electronic medical record system integration: obtain the patient's historical health records through the patient's electronic medical record information, including historical medical history, medication records and surgical records;

[0015] S13, Lifestyle data collection: Obtain patients’ lifestyle data, including daily eating habits, exercise, and sleep quality, through tools such as questionnaires, mobile phone applications, or smart devices;

[0016] S14, environmental parameter collection: Obtain the patient's surrounding environmental parameters through environmental monitoring equipment, including temperature, air quality index and noise level.

[0017] Optionally, S2 includes:

[0018] S21, data denoising: denoising the collected multi-source data to eliminate environmental noise and equipment noise that may exist in the data;

[0019] S22, outlier detection: for the denoised multi-source data, use a statistically based outlier detection method to identify and mark abnormal data points;

[0020] S23, missing value filling: linear interpolation is used to fill missing values ​​in multi-source data after outlier detection;

[0021] S24, data normalization: normalize the filled multi-source data and scale different types of data to the same range;

[0022] S25, timestamp marking: add timestamps to the pre-processed multi-source data and ensure the time consistency of each data source through synchronization technology.

[0023] Optionally, the S3 includes:

[0024] S31, data input: input the preprocessed physiological parameters into the convolutional neural network model;

[0025] S32, convolutional layer construction: applying multiple convolution kernels in the convolutional layer of the convolutional neural network model to identify the characteristic patterns of physiological parameters and extract feature maps of different scales;

[0026] S33, pooling layer processing: adding a pooling layer after the convolution layer to reduce the dimension of physiological parameters through the maximum pooling operation;

[0027] S34, feature extraction: Analyze the physiological parameters after convolution and pooling processing to extract physiological features related to cardiovascular health, including heart rate variability and blood pressure fluctuation trend;

[0028] S35, feature vector generation: integrating the extracted physiological features into a physiological feature vector.

[0029] Optionally, the S3 includes:

[0030] S36, historical health record feature extraction: Based on the patient's historical health records, the principal component analysis method is used to extract historical health feature vectors related to cardiovascular health;

[0031] S37, lifestyle data feature extraction: for lifestyle data, principal component analysis is used to reduce the dimension and extract, and lifestyle feature vectors related to cardiovascular health are identified;

[0032] S38, environmental parameter feature extraction: for environmental parameters, the principal component analysis method is used to extract environmental feature vectors related to cardiovascular health;

[0033] S39, multi-dimensional feature set construction: Integrate the physiological feature vector with the historical health feature vector, lifestyle feature vector and environmental feature vector to form a multi-dimensional feature set.

[0034] Optionally, the S4 includes:

[0035] S41, Random Forest Model Construction: Construct a random forest model containing multiple decision trees to identify cardiovascular health-related risk factors from a multi-dimensional feature set;

[0036] S42, data input: input the multidimensional feature set into the random forest model to perform feature importance analysis and identify risk factors associated with cardiovascular disease;

[0037] S43, Feature importance assessment: In the random forest model, the importance of features in cardiovascular disease prediction was assessed by calculating the contribution of each feature to the splitting node of the decision tree;

[0038] S44, risk factor selection: Based on feature importance, features closely related to cardiovascular health are screened out as the final risk factors.

[0039] Optionally, S4 further includes:

[0040] S45, Logistic regression model construction: Construct a logistic regression model, take the risk factors as independent variables, calculate the regression coefficient of each risk factor, and determine the impact of each risk factor on cardiovascular health risk;

[0041] S46, data input: input each risk factor identified by the random forest model into the logistic regression model, and calculate the weight of each risk factor based on the regression coefficient;

[0042] S47, regression coefficient training: Use the maximum likelihood estimation method to train the logistic regression model, calculate the regression coefficient of each risk factor, and determine its impact weight on the risk of cardiovascular disease;

[0043] S48, generation of individualized risk score: The regression coefficient of each risk factor is multiplied by the eigenvalue and weighted to obtain an individualized risk score for preliminary assessment of the patient's cardiovascular health risk.

[0044] Optionally, the S5 includes:

[0045] S51, long short-term memory network model construction: construct a long short-term memory network model, the model unit structure includes input gate, forget gate, memory unit and output gate;

[0046] S52, data input and preparation: The weighted risk factors are input into the long short-term memory network model in time series so that the model can learn the patterns related to cardiovascular risk in the time series;

[0047] S53, model training: Train the long short-term memory network model through historical multi-source data, and use the back-propagation algorithm to optimize the weight parameters.

[0048] Optionally, S5 further includes:

[0049] S54, real-time data input: input the multi-source data collected in real time into the long short-term memory network model in time series to perform real-time cardiovascular health status analysis;

[0050] S55, dynamic risk score calculation: based on the output of the long short-term memory network model, the risk of cardiovascular events is assessed in real time to generate a dynamic risk score;

[0051] S56, Risk Trend Analysis: As time goes by, new real-time data is continuously input to dynamically generate risk scores, forming a time series trend of the scores. By analyzing the changing trends of risk scores, potential risk increases or decreases can be identified.

[0052] Optionally, the S6 includes:

[0053] S61, Dynamic Risk Score Analysis: Continuously monitor and analyze the dynamic risk score output by the long short-term memory network model to determine the patient's current cardiovascular health status;

[0054] S62, Personalized health intervention recommendation generation: Generate personalized health intervention recommendations based on the patient's dynamic risk score;

[0055] S63, early warning mechanism triggering: If the dynamic risk score exceeds the preset high-risk threshold, the early warning mechanism will be automatically triggered, and an alert will be sent to patients and medical staff through the application to indicate possible urgent health risks;

[0056] S64, Notification and Feedback: When the early warning mechanism is triggered, real-time notifications are automatically sent to the patient’s mobile application or designated medical staff terminal.

[0057] Beneficial effects of the present invention:

[0058] The present invention achieves comprehensive coverage and multi-dimensional acquisition of patient health information through multi-source data collection. Based on convolutional neural networks and principal component analysis, this method can extract features closely related to cardiovascular health from physiological and non-physiological data and fuse them into a multi-dimensional feature set, providing rich input data for cardiovascular disease risk assessment. This multi-source data fusion and precise feature extraction improves the accuracy of cardiovascular risk factor identification, avoids the problem of missing information caused by a single data source, and ensures the reliability of the assessment results.

[0059] The present invention constructs a dynamic prediction model based on long short-term memory network, generates dynamic risk scores in real time through time series analysis of risk factors, and identifies changes in health risks through score trend tracking. This dynamic assessment method based on time series can automatically trigger an early warning mechanism when cardiovascular risk increases, thereby achieving early health risk monitoring. At the same time, random forest algorithm and logistic regression are used to weight each risk factor to generate an individualized risk score, making the assessment results more personalized. Dynamic risk prediction and individualized scoring ensure the continuity and personalization of health monitoring, and improve the timeliness and effectiveness of health intervention.

[0060] The present invention generates personalized health intervention suggestions based on dynamic risk scores, covering lifestyle adjustments, drug management, and regular check-up plans to meet the health needs of patients at different risk levels. The system also integrates automatic early warning and real-time notification mechanisms. When the dynamic risk score exceeds the threshold, the patient and medical staff are notified immediately through the application, making health management proactive and highly responsive. The combination of early warning mechanism and personalized suggestions helps patients take intervention measures early, effectively reducing health risks, while improving the health management efficiency of medical staff, and realizing the full intelligent and personalized management of cardiovascular health. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] 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.

[0062] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0063] Figure 2 Schematic diagram of S3 process of an embodiment of the present invention. DETAILED DESCRIPTION

[0064] 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.

[0065] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0066] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0067] like Figure 1-Figure 2 As shown, the cardiovascular disease data fusion and multi-dimensional health assessment method includes the following steps:

[0068] S1, data collection: by deploying a variety of sensors and wearable devices, and combining with the electronic medical record system to obtain multi-source data of patients, the multi-source data includes physiological parameters, historical health records, lifestyle data and environmental parameters;

[0069] S2, data preprocessing: preprocessing the collected multi-source data, including noise removal, missing value filling, and outlier detection;

[0070] S3, feature extraction and data fusion: Use convolutional neural networks to extract features from physiological parameters and extract features related to cardiovascular health, such as heart rate variability and blood pressure fluctuation trends. Use principal component analysis to extract features from historical health records, lifestyle data, and environmental parameters. Integrate the features extracted from physiological parameters with the features extracted from historical health records, lifestyle data, and environmental parameters to form a multi-dimensional feature set.

[0071] S4, risk factor identification and weighting: The random forest algorithm is used to identify risk factors associated with cardiovascular disease from a multidimensional feature set. At the same time, a logistic regression algorithm is used to assign weights to each risk factor to generate an individualized risk score for preliminary health assessment;

[0072] S5, Predictive modeling and risk assessment: Based on the identified risk factors, a long-short term memory network model is constructed to analyze the time series trends of historical multi-source data and real-time multi-source data, predict the risk of cardiovascular events, and generate dynamic risk scores;

[0073] S6, Personalized advice and intervention: Generate personalized health intervention recommendations based on the dynamic risk score output by the long short-term memory network model, including lifestyle adjustments, drug management, and regular check-up plans. If the dynamic risk score exceeds the preset threshold, the early warning mechanism is automatically triggered to notify patients and medical staff through the application.

[0074] S1 includes:

[0075] S11, Sensor and equipment deployment: Deploy a variety of sensors and wearable devices in the patient's daily living environment to collect physiological parameter data, including electrocardiogram sensors to obtain patients' electrocardiogram signals, blood pressure sensors to monitor patients' blood pressure changes, and heart rate sensors to measure patients' heart rates;

[0076] S12, electronic medical record system integration: obtain the patient's historical health records through the patient's electronic medical record information, including historical medical history, medication records and surgical records;

[0077] S13, Lifestyle data collection: Obtain patients’ lifestyle data, including daily eating habits, exercise, and sleep quality, through tools such as questionnaires, mobile phone applications, or smart devices;

[0078] S14, environmental parameter collection: obtaining the patient's surrounding environmental parameters through environmental monitoring equipment, including temperature, air quality index and noise level;

[0079] By deploying sensors and integrating historical health records, lifestyle, and environmental data, a comprehensive multi-source data collection method is provided, providing detailed and rich data support for cardiovascular health assessment, which lays a solid foundation for subsequent feature extraction and risk assessment.

[0080] S2 includes:

[0081] S21, data denoising: denoising is performed on the collected multi-source data to eliminate the environmental noise and equipment noise that may exist in the data. For physiological parameters, a bandpass filter is used to remove power supply noise (50 / 60Hz) and high-frequency interference to ensure that the data is clear and reliable. The environmental parameter (such as noise level) data is smoothed and filtered to reduce the interference of mutation points;

[0082] S22, outlier detection: For the denoised multi-source data, use a statistically based outlier detection method to identify and mark abnormal data points. The outlier detection formula is as follows:

[0083]

[0084] Among them, Z is the standard score of the data, X is the data value, μ is the mean of the data set, σ is the standard deviation of the data set, and if |Z|>3, it is considered as abnormal data;

[0085] S23, missing value filling: linear interpolation is used to fill missing values ​​for multi-source data after outlier detection. The filling formula is:

[0086]

[0087] Where X t is a missing value, X t-1and X t+1 are the adjacent data points before and after the missing value. For non-time series data (such as lifestyle data), the average value filling method is used;

[0088] S24, data normalization: normalize the filled multi-source data and scale different types of data to the same range. The normalization formula is expressed as:

[0089]

[0090] Among them, X norm is the normalized value, X is the original value, max(X) and min(X) are the maximum and minimum values ​​of the data set, respectively. This step ensures the compatibility of physiological parameters, historical health records, lifestyle data, and environmental parameters in subsequent analyses;

[0091] S25, timestamp tagging: add timestamps to pre-processed multi-source data and ensure the time consistency of each data source through synchronization technology to support time series analysis;

[0092] Through data denoising, outlier detection, missing value filling and data normalization, the integrity, consistency and timeliness of multi-source data are ensured, providing a high-quality data foundation for subsequent feature extraction and risk assessment.

[0093] S3 includes:

[0094] S31, data input: input the preprocessed physiological parameters into the convolutional neural network model;

[0095] S32, convolution layer construction: Multiple convolution kernels are applied in the convolution layer of the convolutional neural network model to identify the characteristic patterns of physiological parameters and extract feature maps of different scales. The convolution calculation formula is:

[0096]

[0097] Among them, f(x) is the output feature, x i is the input data, w i is the convolution kernel weight, b is the bias term, and σ is the activation function (such as ReLU function);

[0098] S33, pooling layer processing: Add a pooling layer after the convolution layer to reduce the dimension of physiological parameters through the maximum pooling operation, which is expressed as:

[0099] y j =max(x j1 ,x j2 ,…,x jn );

[0100] where y jis the feature value after pooling, x j1 ,x j2 ,…,x jn is the original eigenvalue within the pooling window;

[0101] S34, feature extraction: Analyze the physiological parameters after convolution and pooling processing to extract physiological features related to cardiovascular health, including heart rate variability and blood pressure fluctuation trend;

[0102] S35, feature vector generation: integrating the extracted physiological features into a physiological feature vector to characterize the patient's physiological state.

[0103] S3 includes:

[0104] S36, historical health record feature extraction: For the patient's historical health records (such as medical history, medication records, and surgical records), the principal component analysis method is used to extract historical health feature vectors related to cardiovascular health. The principal component analysis method calculates and reduces the dimension through the covariance matrix to extract the main components. The specific formula is expressed as follows:

[0105]

[0106] Where ∑ is the covariance matrix, x i is the data sample, is the data mean, and the main components are retained to constitute the health record characteristics by calculating the eigenvectors and eigenvalues;

[0107] S37, lifestyle data feature extraction: for lifestyle data, principal component analysis is used to reduce the dimension and extract, and lifestyle feature vectors related to cardiovascular health are identified;

[0108] S38, environmental parameter feature extraction: for environmental parameters, the principal component analysis method is used to extract environmental feature vectors related to cardiovascular health. The extracted environmental feature vectors are used to represent the potential impact of environmental factors on cardiovascular health;

[0109] S39, multi-dimensional feature set construction: integrating physiological feature vectors with historical health feature vectors, lifestyle feature vectors, and environmental feature vectors to form a multi-dimensional feature set for subsequent risk factor identification and risk assessment;

[0110] Through principal component analysis, key features of historical health records, lifestyle data, and environmental parameters are extracted, and physiological parameter features are integrated to construct a multidimensional feature set, laying the foundation for accurate cardiovascular risk assessment.

[0111] S4 includes:

[0112] S41, Random Forest Model Construction: Construct a random forest model containing multiple decision trees to identify cardiovascular health-related risk factors from a multi-dimensional feature set. The random forest model improves prediction accuracy and reduces overfitting risk by integrating multiple decision trees. The model formula is expressed as:

[0113]

[0114] Among them, G is the final risk factor score of the model, F(x i ) is the output of each tree, w i is the weight of the tree;

[0115] S42, data input: input the multidimensional feature set into the random forest model to perform feature importance analysis and identify risk factors associated with cardiovascular disease;

[0116] S43, Feature importance evaluation: In the random forest model, the importance of each feature in the prediction of cardiovascular disease is evaluated by calculating the contribution of each feature to the decision tree splitting node. The feature importance evaluation formula is as follows:

[0117]

[0118] Among them, I(f i ) is the feature f i The importance score of N is the total number of trees, T is the set of all decision trees, and I t (f i ) represents the feature f i Node split contribution in tree t;

[0119] S44, risk factor selection: based on feature importance, features closely related to cardiovascular health were selected as the final risk factors;

[0120] The random forest algorithm is used to identify key cardiovascular risk factors in a multi-dimensional feature set, providing a scientific basis for subsequent risk weighting and individualized risk scoring.

[0121] S4 also includes:

[0122] S45, Logistic regression model construction: Construct a logistic regression model, take the risk factor as the independent variable, calculate the regression coefficient of each risk factor, and determine the impact of each risk factor on cardiovascular health risk. The expression of the logistic regression model is:

[0123]

[0124] Among them, p is the individualized risk score, β0 is the bias term, and β i is the regression coefficient of the ith risk factor, f iis the i-th risk factor;

[0125] S46, data input: input each risk factor identified by the random forest model into the logistic regression model, and calculate the weight of each risk factor based on the regression coefficient;

[0126] S47, regression coefficient training: Use the maximum likelihood estimation method to train the logistic regression model, calculate the regression coefficient of each risk factor, and determine its impact weight on the risk of cardiovascular disease. The goal of maximum likelihood estimation is to find the regression coefficient combination that best fits the observed risk factor data;

[0127] S48, Individualized risk score generation: The regression coefficient of each risk factor is multiplied and weighted by the eigenvalue to obtain an individualized risk score, which is used to make a preliminary assessment of the patient's cardiovascular health risk and support subsequent dynamic risk prediction;

[0128] A logistic regression algorithm is used to assign weights to each risk factor and generate an individualized risk score, which provides a preliminary quantitative assessment of the patient's health risks and provides accurate input data for subsequent predictive modeling.

[0129] S5 includes:

[0130] S51, Long Short-Term Memory Network Model Construction: Construct a long short-term memory network model. The model unit structure includes input gate, forget gate, memory unit and output gate to adapt to the long-term dependency of time series data. The long short-term memory network model includes:

[0131] (1) Forget Gate:

[0132] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0133] Among them, f t is the output of the forget gate, W f is the weight matrix of the forget gate, h t-1 is the hidden state of the previous time step, x t is the current input, b f is the bias term, σ is the Sigmoid activation function;

[0134] (2) Input gate:

[0135] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0136]

[0137] Among them, i t is the output of the input gate, W i is the input gate weight, is the candidate memory cell value, W C is the candidate memory unit weight matrix, b i and b C is the bias term;

[0138] (3) Memory unit update:

[0139]

[0140] Among them, C t is the current memory cell state, C t-1 is the memory cell state at the previous time step;

[0141] (4) Output gate:

[0142] o t =σ(W o ·[h t-1 ,x t ]+b o );

[0143] h t =o t ×tanh(C t );

[0144] Among them, t is the output of the output gate, W o is the output gate weight, h t is the hidden state of the current time step;

[0145] S52, data input and preparation: The weighted risk factors are input into the long short-term memory network model in time series so that the model can learn the patterns related to cardiovascular risk in the time series;

[0146] S53, model training: The long short-term memory network model is trained through historical multi-source data, and the weight parameters are optimized using the back propagation algorithm, so that the model gradually learns and captures the time series characteristics of the data, thereby improving the accuracy of the prediction;

[0147] Through the structural design and training process of the long short-term memory network model, a dynamic prediction model for cardiovascular risk was established, providing a basis for subsequent risk assessment and dynamic score generation.

[0148] The S5 also includes:

[0149] S54, real-time data input: input the multi-source data (including physiological parameters, lifestyle data and environmental parameters) collected in real time into the long short-term memory network model in time series to perform real-time cardiovascular health status analysis;

[0150] S55, dynamic risk score calculation: Based on the output of the long short-term memory network model, the risk of cardiovascular events is evaluated in real time to generate a dynamic risk score. The dynamic risk score is calculated using the following formula

[0151] Dynamic risk score = h t

[0152] where h t is the output value of the long short-term memory network at the current time step, indicating the instant score of cardiovascular risk prediction;

[0153] S56, Risk Trend Analysis: As time goes by, new real-time data is continuously input to dynamically generate risk scores and form time series trends of the scores. By analyzing the changing trends of risk scores, potential rising or falling risks can be identified to provide data support for subsequent health interventions and management;

[0154] Through real-time data analysis, risk score calculation and trend monitoring of the long short-term memory network model, dynamic monitoring of the patient's cardiovascular health status is achieved, providing real-time support for early warning and personalized intervention.

[0155] S6 includes:

[0156] S61, Dynamic risk score analysis: Continuously monitor and analyze the dynamic risk score output by the long short-term memory network model to determine the patient's current cardiovascular health status and determine the corresponding intervention needs based on the different levels of the score (such as low, medium, and high risk);

[0157] S62, Personalized health intervention recommendation generation: Generate personalized health intervention recommendations based on the patient's dynamic risk score, including:

[0158] Lifestyle modification recommendations: Based on the patient’s risk level, appropriate diet, exercise, and sleep modification plans are recommended. For example, high-risk patients may receive stricter dietary control and exercise restriction recommendations.

[0159] Medication management recommendations: Based on the patient's dynamic risk score and medication records, medication management recommendations are provided. For patients with high risk scores, it is recommended to regularly review medication effects and adjust medication dosages as appropriate;

[0160] Regular check-up plan: For patients at medium and high risk, develop and recommend regular cardiovascular health check-up plans, including regular monitoring of electrocardiogram, blood pressure, blood lipids and other indicators, so as to continuously track health changes;

[0161] S63, early warning mechanism triggering: If the dynamic risk score exceeds the preset high-risk threshold, the early warning mechanism will be automatically triggered, and an alert will be sent to patients and medical staff through the application to indicate possible urgent health risks;

[0162] S64, Notification and Feedback: When the early warning mechanism is triggered, a real-time notification is automatically sent to the patient’s mobile application or designated medical staff terminal. The notification content includes the current dynamic risk score, recommended emergency intervention measures, and further health management suggestions.

[0163] 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.

[0164] 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. Cardiovascular disease data fusion and multi-dimensional health assessment method, characterized by: The following steps are involved: S1, data collection: by deploying a variety of sensors and wearable devices, and combining with the electronic medical record system to obtain multi-source data of patients, the multi-source data includes physiological parameters, historical health records, lifestyle data and environmental parameters; S2, data preprocessing: preprocessing the collected multi-source data, including noise removal, missing value filling, and outlier detection; S3, feature extraction and data fusion: use convolutional neural network to extract features from physiological parameters, use principal component analysis to extract features from historical health records, lifestyle data and environmental parameters, integrate the features extracted from physiological parameters with the features extracted from historical health records, lifestyle data and environmental parameters to form a multi-dimensional feature set; S4, risk factor identification and weighting: The random forest algorithm is used to identify risk factors associated with cardiovascular disease from a multidimensional feature set. At the same time, a logistic regression algorithm is used to assign weights to each risk factor to generate an individualized risk score for preliminary health assessment; S5, Predictive modeling and risk assessment: Based on the identified risk factors, a long-short term memory network model is constructed to analyze the time series trends of historical multi-source data and real-time multi-source data, predict the risk of cardiovascular events, and generate dynamic risk scores; S6, Personalized advice and intervention: Generate personalized health intervention recommendations based on the dynamic risk score output by the long short-term memory network model, including lifestyle adjustments, drug management, and regular check-up plans. If the dynamic risk score exceeds the preset threshold, the early warning mechanism is automatically triggered to notify patients and medical staff through the application.

2. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 1, characterized in that: The S1 includes: S11, Sensor and equipment deployment: Deploy a variety of sensors and wearable devices in the patient's daily living environment to collect physiological parameter data, including electrocardiogram sensors to obtain patients' electrocardiogram signals, blood pressure sensors to monitor patients' blood pressure changes, and heart rate sensors to measure patients' heart rates; S12, electronic medical record system integration: obtain the patient's historical health records through the patient's electronic medical record information, including historical medical history, medication records and surgical records; S13, Lifestyle data collection: Obtain patients’ lifestyle data, including daily eating habits, exercise, and sleep quality, through tools such as questionnaires, mobile phone applications, or smart devices; S14, environmental parameter collection: Obtain the patient's surrounding environmental parameters through environmental monitoring equipment, including temperature, air quality index and noise level.

3. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 2, characterized in that: The S2 includes: S21, data denoising: denoising the collected multi-source data to eliminate environmental noise and equipment noise that may exist in the data; S22, outlier detection: for the denoised multi-source data, use a statistically based outlier detection method to identify and mark abnormal data points; S23, missing value filling: linear interpolation is used to fill missing values ​​in multi-source data after outlier detection; S24, data normalization: normalize the filled multi-source data and scale different types of data to the same range; S25, timestamp marking: add timestamps to the pre-processed multi-source data and ensure the time consistency of each data source through synchronization technology.

4. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 3, characterized in that: The S3 includes: S31, data input: input the preprocessed physiological parameters into the convolutional neural network model; S32, convolutional layer construction: applying multiple convolution kernels in the convolutional layer of the convolutional neural network model to identify the characteristic patterns of physiological parameters and extract feature maps of different scales; S33, pooling layer processing: adding a pooling layer after the convolution layer to reduce the dimension of physiological parameters through the maximum pooling operation; S34, feature extraction: Analyze the physiological parameters after convolution and pooling processing to extract physiological features related to cardiovascular health, including heart rate variability and blood pressure fluctuation trend; S35, feature vector generation: integrating the extracted physiological features into a physiological feature vector.

5. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 4, characterized in that: The S3 includes: S36, historical health record feature extraction: Based on the patient's historical health records, the principal component analysis method is used to extract historical health feature vectors related to cardiovascular health; S37, lifestyle data feature extraction: for lifestyle data, principal component analysis is used to reduce the dimension and extract, and lifestyle feature vectors related to cardiovascular health are identified; S38, environmental parameter feature extraction: for environmental parameters, the principal component analysis method is used to extract environmental feature vectors related to cardiovascular health; S39, multi-dimensional feature set construction: Integrate the physiological feature vector with the historical health feature vector, lifestyle feature vector and environmental feature vector to form a multi-dimensional feature set.

6. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 5, characterized in that: The S4 includes: S41, Random Forest Model Construction: Construct a random forest model containing multiple decision trees to identify cardiovascular health-related risk factors from a multi-dimensional feature set; S42, data input: input the multidimensional feature set into the random forest model to perform feature importance analysis and identify risk factors associated with cardiovascular disease; S43, Feature importance assessment: In the random forest model, the importance of features in cardiovascular disease prediction was assessed by calculating the contribution of each feature to the splitting node of the decision tree; S44, risk factor selection: Based on feature importance, features related to cardiovascular health are screened out as the final risk factors.

7. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 6, characterized in that: The S4 further comprises: S45, Logistic regression model construction: Construct a logistic regression model, take the risk factors as independent variables, calculate the regression coefficient of each risk factor, and determine the impact of each risk factor on cardiovascular health risk; S46, data input: input each risk factor identified by the random forest model into the logistic regression model, and calculate the weight of each risk factor based on the regression coefficient; S47, regression coefficient training: Use the maximum likelihood estimation method to train the logistic regression model, calculate the regression coefficient of each risk factor, and determine its impact weight on the risk of cardiovascular disease; S48, generation of individualized risk score: The regression coefficient of each risk factor is multiplied by the eigenvalue and weighted to obtain an individualized risk score for preliminary assessment of the patient's cardiovascular health risk.

8. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 7, characterized in that: The S5 includes: S51, long short-term memory network model construction: construct a long short-term memory network model, the model unit structure includes input gate, forget gate, memory unit and output gate; S52, data input and preparation: The weighted risk factors are input into the long short-term memory network model in time series so that the model can learn the patterns related to cardiovascular risk in the time series; S53, model training: Train the long short-term memory network model through historical multi-source data, and use the back-propagation algorithm to optimize the weight parameters.

9. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 8, characterized in that: The S5 further includes: S54, real-time data input: input the multi-source data collected in real time into the long short-term memory network model in time series to perform real-time cardiovascular health status analysis; S55, dynamic risk score calculation: based on the output of the long short-term memory network model, the risk of cardiovascular events is assessed in real time to generate a dynamic risk score; S56, Risk Trend Analysis: As time goes by, new real-time data is continuously input to dynamically generate risk scores, forming a time series trend of the scores. By analyzing the changing trends of risk scores, potential risk increases or decreases can be identified.

10. The cardiovascular disease data fusion and multi-dimensional health assessment method according to claim 9, characterized in that: The S6 includes: S61, Dynamic Risk Score Analysis: Continuously monitor and analyze the dynamic risk score output by the long short-term memory network model to determine the patient's current cardiovascular health status; S62, Personalized health intervention recommendation generation: Generate personalized health intervention recommendations based on the patient's dynamic risk score; S63, early warning mechanism triggering: If the dynamic risk score exceeds the preset high-risk threshold, the early warning mechanism will be automatically triggered, and an alert will be sent to the patient and medical staff through the application; S64, Notification and Feedback: When the early warning mechanism is triggered, real-time notifications are automatically sent to the patient’s mobile application or designated medical staff terminal.

Images