Omnibearing non-invasive heart failure monitoring system and monitoring method

Through multi-parameter wearable devices and cloud processors combined with machine learning algorithms, we can monitor and analyze the physiological parameters of heart failure patients in real time, provide personalized health management suggestions, and solve the problem of lack of deep learning and user interface unfriendly in the existing technology. Improves the management efficiency and quality of life of heart failure.

CN119033345APending Publication Date: 2024-11-29PEKING UNION MEDICAL COLLEGE HOSPITAL +1
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411164289.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2024-07-03
Filing Date
2024-08-23
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The existing heart failure monitoring systems lack the long-term trend of deep learning or artificial intelligence algorithms to analyze and predict patient status, and the user interface is unfriendly, making it difficult to provide personalized medical advice and rehabilitation plans, limiting the long-term health management of patients.

Method used

Using multi-parameter wearable devices to upload data in real time to the cloud processor through wireless communication, using machine learning algorithms such as ET-ResNet, RNN and LSTM models for data analysis, providing personalized health management suggestions, and through user-friendly mobile application layer Support instant communication between doctors and patients.

Benefits of technology

Accurate identification and risk assessment of early signs of heart failure, provide personalized health management advice, improve patients' quality of life and treatment effects, and reduce medical costs and operational complexity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119033345A_ABST
    Figure CN119033345A_ABST
Patent Text Reader

Abstract

The invention discloses an omnibearing non-intrusive heart failure monitoring system and a monitoring method.The monitoring system comprises a multi-parameter monitoring wearable device, a cloud processor and an application layer, and monitoring data collected by the multi-parameter monitoring wearable device is uploaded to the cloud processor in real time through wireless communication; the cloud processor predicts and analyzes the collected monitoring data through a machine learning algorithm, and the patient and the doctor check the monitoring data in real time through the application layer; the monitoring method comprises the following steps that the multi-parameter monitoring wearable device uploads continuously monitored heart rate, thoracic impedance, activity state and electrocardiogram to a cloud database; the cloud processor predicts the heart failure risk and identifies early signs of heart failure; and the application layer provides personalized health management suggestions according to the processing result of the cloud processor. According to the invention, the physiological parameters of the heart failure patient can be continuously monitored in real time, the early symptoms of heart failure are identified, and the management efficiency of chronic heart failure is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of medical health monitoring technology, and in particular to an all-round non-invasive heart failure monitoring system and monitoring method. Background Art

[0002] Heart failure is a serious heart disease characterized by a decrease in the heart's pumping function that requires continuous monitoring and management. Existing heart failure monitoring methods mostly rely on regular medical examinations or invasive monitoring devices, which are not only costly but also disruptive to patients' daily lives. With the advancement of technology, the demand for non-invasive and continuous monitoring technologies is increasing.

[0003] For example, CN115844442A shows a system that uses posture and sound sensors to monitor patients with heart failure. The system can issue an early warning by analyzing changes in lung and respiratory sounds when the patient changes posture, such as from sitting to lying. This technology is particularly suitable for tracking the daily activities and physiological status of patients with chronic heart failure, and improves the ability to monitor the early stage symptoms of chronic left heart failure.

[0004] CN115662629A describes a heart failure monitoring system that uses portable sensor devices to collect heart failure monitoring data and analyzes the data through a heart failure monitoring management system. The system uses the HOHF algorithm model, combined with human body impedance and limb lead ECG data, as well as weight and other biological parameters to predict and evaluate the risk of heart failure. The system is designed to monitor heart failure conditions in real time and provide effective early warnings, reduce the re-hospitalization rate of heart failure patients, and improve the quality of life of patients.

[0005] CN114983355A relates to a heart failure monitoring system, which includes a heart monitor and a signal acquisition module and processor for acquiring and processing heart failure-related physiological signals. This system focuses on real-time monitoring and determines heart failure indicators through preset calculation rules, which helps doctors make more accurate disease diagnosis and treatment adjustments. Nevertheless, this system still has shortcomings in personalized processing and technical maintenance, and the technical complexity may limit its application in non-professional environments.

[0006] Although existing technologies have improved the efficiency of chronic heart failure monitoring to a certain extent, there are still some limitations. Existing systems mostly rely on direct prompts from sensor data and lack deep learning or artificial intelligence algorithms to analyze and predict long-term trends in patient status. Although existing systems are capable of monitoring and early warning, they generally do not provide personalized medical advice or rehabilitation programs, which are essential for long-term health management of patients. In addition, existing solutions are often not user-friendly enough in terms of user interface and interaction, especially for elderly patients, who may encounter challenges in use. Summary of the invention

[0007] Purpose of the invention: The purpose of the present invention is to provide a comprehensive non-invasive heart failure monitoring system and monitoring method, which can update its prediction model in real time to adapt to changes in the patient's condition, provide personalized health management suggestions, effectively improve the patient's quality of life and reduce medical costs.

[0008] Technical solution: A comprehensive non-invasive heart failure monitoring system, including a multi-parameter monitoring wearable device, a cloud processor and an application layer. The monitoring data collected by the multi-parameter monitoring wearable device is uploaded to the cloud processor in real time through wireless communication. The cloud processor predicts and analyzes the collected monitoring data through machine learning algorithms. Patients and doctors view the monitoring data in real time through the application layer.

[0009] The multi-parameter monitoring wearable device can continuously monitor heart rate, chest impedance, activity status and electrocardiogram;

[0010] The cloud-based processor can predict the risk of heart failure and identify early signs of heart failure;

[0011] The application layer can provide personalized health management suggestions and support instant communication between doctors and patients.

[0012] Further, the multi-parameter monitoring wearable device includes a heart rate and electrocardiogram monitoring module, a respiration monitoring module, a chest impedance measurement module, an activity status monitoring module, a microcontroller, and a Wi-Fi module;

[0013] The heart rate and electrocardiogram monitoring module uses photoplethysmography to measure blood flow changes and monitor heart rate by emitting and receiving light waves that penetrate the skin and blood vessels;

[0014] The respiratory monitoring module detects inhalation or exhalation by measuring the chest pressure changes caused by respiratory movement, providing reliable data for respiratory monitoring;

[0015] The thoracic impedance measurement module monitors potential fluid accumulation in heart failure patients by measuring impedance changes in the thoracic area;

[0016] The activity status monitoring module monitors the patient's activity status by measuring the acceleration changes in three directions to identify the patient's resting, walking and other activity modes;

[0017] The microcontroller collects the patient's physiological data in real time;

[0018] All monitoring data is uploaded to the cloud database in real time via the built-in Wi-Fi module. The cloud processor conducts in-depth analysis of the collected data by identifying and analyzing heart failure models to predict heart failure events and assess heart failure risks.

[0019] The monitoring method of the above-mentioned all-round non-invasive heart failure monitoring system is implemented in the following steps:

[0020] S1, wear the multi-parameter wearable device on the designated part to ensure good contact between the sensor and the skin;

[0021] S2, after the device is started, the sensor collects physiological parameters in real time; the collected data is preliminarily processed by the device's built-in microcontroller and uploaded to the cloud database in real time stably and quickly through the optimized Wi-Fi module;

[0022] S3, filtering, normalizing and feature extraction of the raw data uploaded to the cloud database;

[0023] S4, extract key features from various physiological parameters through the extreme tree and perform preliminary classification; and input the features output by the extreme tree into the residual neural network;

[0024] S5, using RNN and LSTM models to process time series data, capture physiological states that evolve over time, and identify early signs of heart failure;

[0025] S6, comprehensively analyzing the patient's physiological data by identifying and analyzing the heart failure model to assess the risk level of heart failure;

[0026] S7, when the monitoring system detects early signs of heart failure, it automatically generates warning information and notifies patients and doctors through mobile applications; and generates personalized health management suggestions based on the patient's specific condition and analysis results;

[0027] S8, the monitoring system continuously monitors the patients and regularly updates the rehabilitation plan; based on the latest data analysis and patient feedback, the rehabilitation recommendations are adjusted to ensure the effectiveness and adaptability of the rehabilitation program.

[0028] Furthermore, heart rate is measured by photoplethysmography, ECG measures the heart's electrical activity through electrodes, chest impedance is measured by tiny currents, and activity status is monitored by an accelerometer.

[0029] Furthermore, the pre-processed data is presented in the form of graphs and numerical values ​​in a user-friendly mobile application, allowing patients and doctors to intuitively view real-time and historical data;

[0030] Patients and doctors can understand their health status and view personalized health management recommendations through the user interface.

[0031] Further, in step S6, the monitoring system regularly updates the recognition and analysis model of heart failure to adapt to changes in the patient's condition and ensure the accuracy of the prediction;

[0032] The implementation process of building a model for identifying and analyzing heart failure is as follows:

[0033] S61, continuously collects the patient's physiological data through sensors integrated into multi-parameter wearable devices; S62,

[0034] The collected physiological data are preprocessed and key features are extracted from the raw data;

[0035] S63, using the ET-ResNet model to perform classification and regression analysis by evaluating the importance of individual physiological parameters and outputting prediction results;

[0036] S64, regularly adjust the parameters of the ET-ResNet model based on the evaluation and feedback of the prediction results.

[0037] Furthermore, the warning information includes current physiological parameters, risk level, and intervention measures;

[0038] Depending on the level of risk, the surveillance system provides the following interventions:

[0039] Low risk: Patients are advised to increase the frequency of daily monitoring and maintain a healthy lifestyle;

[0040] Medium risk: Patients are advised to consult a doctor and may need to adjust medication dosage or undergo further examination;

[0041] High risk: Immediately notify the doctor and family members and advise the patient to go to the hospital for emergency examination and treatment.

[0042] Compared with the prior art, the present invention has the following significant effects:

[0043] 1. The monitoring system of the present invention adopts a machine-deep ET-ResNet architecture and also utilizes RNN, LSTM and multi-layer perceptron (MLP) models to achieve more accurate risk assessment and more effective data processing; the system can also update its prediction model in real time to adapt to changes in the patient's condition, provide personalized health management suggestions, effectively improve the patient's quality of life and reduce medical costs, thereby solving key problems in the prior art;

[0044] 2. The monitoring system of the present invention can continuously monitor the physiological parameters of heart failure patients in real time, including heart rate, chest impedance, activity status and electrocardiogram (ECG). Through comprehensive monitoring, it can timely detect and warn of signs of worsening heart failure. Combined with deep learning algorithms, the system can identify early symptoms of heart failure, significantly improve the management efficiency of chronic heart failure, and provide doctors with accurate data support to make timely treatment decisions. Through early diagnosis and timely intervention, the treatment effect and survival rate of patients with heart failure can be improved;

[0045] 3. The present invention adopts a user-friendly interactive interface to ensure that patients and doctors can easily access and manage health data, and can achieve efficient and safe monitoring even under the operation of non-professionals, thereby improving the patient's experience and reducing the complexity of operation;

[0046] 4. The monitoring method of the present invention customizes personalized health management suggestions according to the patient's specific conditions, thereby improving the pertinence and effectiveness of treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0048] Figure 2 A flowchart for the overall data flow and information processing from the multi-parameter monitoring wearable device to the cloud database and then to the mobile application;

[0049] Figure 3 Schematic diagram for building a model for identifying and analyzing heart failure;

[0050] Figure 4 The figure is a schematic diagram of the implementation process of a heart failure monitoring method of the present invention. DETAILED DESCRIPTION

[0051] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0052] like Figure 1 , Figure 2 As shown, the present invention provides a comprehensive non-invasive heart failure monitoring system, which aims to identify signs of heart failure in advance by real-time monitoring and analysis of key physiological parameters, and timely intervention to optimize the treatment effect. It mainly includes the following contents:

[0053] Multi-parameter monitoring wearable device: A wearable device that integrates multiple physiological sensors has been developed to continuously monitor heart rate, thoracic impedance, activity status, and electrocardiogram (ECG). These parameters are critical for real-time assessment of heart failure conditions. For example, a decrease in thoracic impedance may indicate fluid accumulation, which is an early indicator of worsening heart failure.

[0054] Internet of Things (IoT) technology: Data collected by wearable devices is uploaded to a cloud database in real time via Wi-Fi technology. This setup allows medical providers to access critical physiological data in real time via the Internet, allowing them to make fast and accurate treatment decisions.

[0055] Cloud processor: Uses advanced machine learning techniques, such as ET-ResNet architecture, RNN, LSTM, and multi-layer perceptron (MLP) models, to analyze large amounts of collected data to accurately predict heart failure risk and identify early signs of heart failure. Figure 3 The data processing flow chart is shown, including the entire process of data collection, preprocessing, feature extraction and model prediction. These algorithms are able to process complex data patterns and learn from them the patient's unique health trajectory.

[0056] User-friendly mobile application layer: A comprehensive mobile application has been developed to enable patients and doctors to view monitoring data in real time and provide personalized health management suggestions. The application also includes a doctor-patient communication function, supporting instant communication between doctors and patients, providing patients with more personalized and immediate medical support.

[0057] 1. Design and function of multi-parameter monitoring wearable devices

[0058] (11) Heart rate and electrocardiogram (ECG) monitoring module

[0059] The heart rate and electrocardiogram (ECG) monitoring module uses photoplethysmography (PPG) technology to measure blood flow changes to monitor heart rate by emitting and receiving light waves that penetrate the skin and blood vessels. Electrocardiogram (ECG) monitoring generates ECG waveforms by amplifying and filtering the heart's electrical activity signals, which is essential for monitoring heart health. The raw PPG and ECG signals are digitized by the analog-to-digital converter (ADC) in the microcontroller, and digital signal processing techniques such as filtering and denoising are applied to improve signal quality. The heart rate is calculated by analyzing the time interval between each heartbeat in the processed PPG signal, while the analysis of the ECG signal includes R-wave detection for calculating heart rate and analyzing heart rhythm.

[0060] (12) Respiratory monitoring module

[0061] The module utilizes laser-induced graphene (LIG), a 3D porous graphene material that was chosen for its simple synthesis process, controllable design and morphology, and excellent electrical and mechanical properties. Laser irradiation produces copper traces on the polyimide (PI) layer, forming a stable electrical connection between the LIG sensing area. A single PI film covering the top of the LIG area forms another layer of LIG. In the last step of the manufacturing process, the sensor is coated with a layer of elastic silicon to enhance its biocompatibility and protect it from external damage. The sensor detects inhalation or exhalation by measuring changes in chest pressure caused by respiratory movement, providing reliable data for respiratory monitoring.

[0062] (13) Thoracic impedance measurement module

[0063] Thoracic impedance measurement relies on passing tiny electrical currents through the skin and measuring the resistance these currents encounter. Congestion in the heart or accumulation of fluid in the lungs can cause changes in chest impedance. This module monitors heart failure patients for potential fluid accumulation, a key indicator of worsening heart failure, by measuring impedance changes in the chest area. A decrease in chest impedance indicates excess fluid and requires immediate medical intervention.

[0064] (14) Activity status monitoring module

[0065] This module uses an accelerometer to monitor the patient's activity status by measuring changes in acceleration in three directions. These data are analyzed through time windows, activity intensity and duration are calculated, and automatically classified by algorithms to identify the patient's resting, walking, and other activity patterns.

[0066] (15) Microcontroller and Wi-Fi module

[0067] The system has a built-in high-performance microcontroller that is responsible for collecting and processing sensor data, optimizing power consumption during measurement tasks, and entering an idle state when no measurement is being performed. The microcontroller collects the patient's physiological data in real time and uploads it to the cloud database in real time through the built-in Wi-Fi module. In the cloud, the collected data is deeply analyzed by identifying and analyzing heart failure models to predict heart failure events and assess heart failure risks. These algorithms are able to learn from large amounts of historical and real-time data, identify key patterns and trends, and provide accurate clinical decision support.

[0068] (II) Constructing a model for identifying and analyzing heart failure

[0069] In the present invention, the core of the design of cloud database and data processing technology is to ensure real-time, accurate and secure data processing and storage, and provide continuous monitoring and analysis for patients with heart failure. Figure 3 As shown in the figure, the implementation process of building a model for identifying and analyzing heart failure is as follows:

[0070] (21) Data collection and preprocessing

[0071] First, the patient's physiological data is continuously collected through sensors integrated in the wearable device, such as heart rate monitor, electrocardiogram (ECG), chest impedance and activity monitor. These data are collected in real time by the microcontroller and uploaded to the cloud database through the Wi-Fi module.

[0072] Preprocessing steps: Filtering: Apply a low-pass filter to the ECG signal to remove noise.

[0073] Normalization: All physiological parameters are normalized to ensure the consistency of model input.

[0074] Feature extraction: Extract key features from the raw data, such as heart rate variability, QRS complex wave characteristics, etc.

[0075] (22) Model construction and training

[0076] In this embodiment, a comprehensive deep learning framework is designed, combining the ET-ResNet architecture with RNN, LSTM and MLP models to analyze and predict heart failure. The construction and function of each model component are described in detail below, including specific implementation methods and formulas.

[0077] (221) ET-ResNet architecture:

[0078] ET-ResNet is an integrated learning model that combines the characteristics of extreme trees (Trees) and residual neural networks (ResNet). The model first uses multiple decision trees to extract features and perform preliminary classification, and then further processes the output of the decision trees through ResNet.

[0079] Extreme Tree Feature Extraction: Extreme trees are used to independently extract key features and health indicators from various physiological parameters such as heart rate and ECG. These extreme trees perform classification and regression analysis by evaluating the importance of individual variables to make a preliminary judgment on the health status of the patient.

[0080] Each extreme tree independently processes a specific physiological signal input X and outputs a set of features T i (X), where T i (X) represents the output feature vector after the i-th limit tree processes the input X, where X is the input physiological parameter, such as heart rate data or ECG signal.

[0081] Integrate features into ResNet:

[0082] The purpose of ensemble learning is to merge the outputs of multiple extreme trees and use ResNet to further extract and learn the complex relationships and patterns between these features to enhance the accuracy of prediction.

[0083] ResNet structure details: ResNet deeply analyzes the integrated features through multiple residual blocks, which can effectively identify and classify potential signs of heart failure.

[0084] y=ResNet(f1(T1(X)),f2(T2(X)),...,f n (T n (X)))

[0085] ResNet configuration: The construction contains multiple residual blocks, each of which consists of two convolutional layers and a skip connection. The ReLU activation function is used to handle nonlinear relationships, and the softmax function is used in the final layer to output the classification probability.

[0086] Through the ET-ResNet architecture, the present invention can fully utilize the advantages of extreme trees in feature selection and the ability of ResNet in pattern recognition to achieve accurate prediction of heart failure symptoms. The ET-ResNet architecture not only improves the accuracy of prediction, but also enables the model to process and analyze large-scale physiological data, which is suitable for complex medical prediction tasks.

[0087] (222) Integrated application of RNN and LSTM

[0088] In an embodiment, RNN and LSTM networks are used to process and analyze time series data, especially electrocardiogram (ECG) and heart rate data. RNN and LSTM networks are particularly suitable for processing physiological signals with time dependence and can capture physiological states that evolve over time, which is of great significance for identifying early signs of heart failure.

[0089] Sequence data processing: The core function of RNN and LSTM is to process sequence data and understand the temporal correlation in the data, which is especially important for continuously monitored physiological parameters such as ECG and heart rate variability.

[0090] RNNs and LSTMs track historical information by maintaining an internal state, enabling them to predict or identify patterns and trends based on time series.

[0091] Characteristics of LSTM network: LSTM is an improved version of RNN, which overcomes the short-term memory problem of traditional RNN when processing long sequence data through a specific gating mechanism.

[0092] The LSTM unit consists of three main gating structures: forget gate, input gate, and output gate. These gating structures help the model avoid the problem of gradient disappearance while maintaining long-term dependencies.

[0093] Forget Gate: Determines which information should be discarded from the cell state.

[0094] f t =σ(W f *[h t-1 ,x t ]+b f )

[0095] Among them, f t is the output of the forget gate, σ is the sigmoid activation function, Wf is the weight matrix of the forget gate, h t-1 is the output of the previous time step, x t is the input of the current time step, b f is the bias term of the forget gate.

[0096] Input Gate: Determines what new information is stored in the cell state.

[0097] i t =σ(W i *[h t-1 ,x t ]+b i )

[0098] Among them, i t is the output of the input gate, W i is the weight matrix of the input gate, b i is the bias term of the input gate.

[0099] Cell state update: Update the cell state, which is the "memory" part in LSTM, combining the information from the forget gate and the input gate.

[0100] c t =f t *c t-1 +i t *tanh(W c *[h t-1 ,x t ]+b c )

[0101] Among them, c t is the current cell state, c t-1 is the cell state at the previous time step, W c is the weight matrix of the unit state, b c is the bias term.

[0102] Output Gate: Determines the value of the next hidden state. The hidden state contains useful information about the previous state and is used for prediction or other decision-making processes.

[0103] h t =o t *tanh(c t )

[0104] o t =σ(W o ·[h t-1 ,x t ]+b o ) is the output of the output gate, W o is the weight matrix of the output gate, bo is the bias term of the output gate.

[0105] Through this detailed configured LSTM model, the present invention is able to effectively process and analyze ECG and heart rate data with long-term dependencies, providing deep insights into the development of heart failure symptoms.

[0106] (223)MLP structure and voting classifier:

[0107] In the study of heart failure prediction, the combination of multi-layer perceptron (MLP) structure and voting classifier is used to improve the prediction accuracy and robustness of the model.

[0108] MLP structure configuration

[0109] Network configuration: MLP consists of multiple fully connected layers, each of which uses the ReLU activation function to increase the nonlinear processing capability of the model.

[0110] Avoid overfitting: Use dropout technology in the hidden layer, usually setting the dropout rate to 0.2 to randomly discard some network connections and enhance the generalization ability of the model.

[0111] The final output layer adopts a softmax activation function, which converts the network output into a probability distribution, which helps in determining the predicted probability of each class in multi-classification problems.

[0112] y M =softmax(W k *relu(...relu(W2*relu(W1*x+b1)+b2)...)+b k )

[0113] Among them, y M represents the final output of the MLP structure, x represents the data of the input layer, W1, W2, ..., W k are the weights of each layer, b1,b2,...,b k is the bias of each layer, and relu is the activation function used to add nonlinearity to handle more complex data relationships.

[0114] Application of voting classifier

[0115] Function and purpose: The voting classifier makes the final decision by combining the prediction results of multiple models. This method can balance the weaknesses of each model and enhance the stability and accuracy of the overall prediction.

[0116] Combining the results of Extreme Tree (ET) and Residual Neural Network (ResNet), each classifier predicts the probability of the category, and these probabilities are weighted averaged according to the set weights to obtain the final category probability. It is suitable for situations where the predicted probability is available and is generally considered to be a more sophisticated and flexible voting method.

[0117] Final Prediction = argmax(∑ n i=1 w i ·p i )

[0118] Among them, p i is the predicted probability of the ith model, w i is the weight of the corresponding model, n is the number of models, and the argmax() function is used to select the category with the highest weighted average probability.

[0119] By combining MLP and voting classifier, this embodiment provides a powerful and flexible method to predict heart failure, which significantly improves the accuracy and reliability of the model in predicting heart failure by utilizing the deep learning capability of MLP and the ensemble learning advantage of voting classifier.

[0120] 3. Heart failure monitoring methods

[0121] like Figure 4 As shown, the present invention provides a heart failure monitoring method, which uses multi-parameter wearable devices, data processing technology and personalized health management solutions to monitor the physiological conditions of heart failure patients in real time and continuously, and provide personalized rehabilitation suggestions. The specific implementation process is as follows:

[0122] (31) Use of multi-parameter wearable devices

[0123] Wearable devices equipped with multi-parameter sensors can continuously monitor heart rate, electrocardiogram (ECG), chest impedance and activity status. The use of wearable devices includes:

[0124] (311) Equipment Wearing

[0125] Patients wear multi-parameter wearable devices on designated parts, such as the wrist or chest, to ensure good contact between the sensor and the skin. The device design should focus on comfort and stability to ensure long-term wearing comfort and sensor accuracy.

[0126] (312) Data Collection

[0127] After the device is started, the sensor collects physiological parameters in real time. The heart rate is measured by photoplethysmography (PPG), the ECG measures the heart's electrical activity through electrodes, the chest impedance is measured by a tiny current, and the activity status is monitored by an accelerometer.

[0128] (313) Data transmission

[0129] The collected data is preliminarily processed by the device’s built-in microcontroller and uploaded to the cloud database in real time and stably and quickly through the optimized Wi-Fi module to ensure the reliability of data transmission.

[0130] (32) Data display method

[0131] (321) Data Preprocessing

[0132] The raw data uploaded to the cloud database undergoes filtering, normalization, and feature extraction to improve data quality and consistency.

[0133] (322) Data Visualization

[0134] The pre-processed data is presented in the form of charts and numerical values ​​in a user-friendly mobile application, where patients and doctors can intuitively view real-time and historical data.

[0135] (323) User Interface

[0136] The app provides a simple and intuitive user interface that allows patients and doctors to quickly understand their health status, set warning thresholds, and view personalized health management recommendations.

[0137] (33) Heart failure monitoring methods

[0138] (331) Model analysis and prediction

[0139] (3311) Feature selection and preliminary classification

[0140] Key features were extracted from various physiological parameters by extreme tree (ET) and preliminary classification was performed.

[0141] (3312) Deep Feature Extraction

[0142] The features output by the extreme tree are input into the residual neural network (ResNet), and the complex relationships and patterns between the features are further extracted and learned through multiple residual blocks.

[0143] (3313) Time Series Analysis

[0144] The time series data is processed using RNN and LSTM models to capture the physiological state that evolves over time and identify early signs of heart failure, including abnormal fluctuations in heart rate, changes in chest impedance, and ECG features such as changes in QRS waveforms.

[0145] (3314) Adaptive Model Update

[0146] The system regularly updates the heart failure identification and analysis model to adapt to changes in patient conditions and ensure the accuracy of predictions.

[0147] (332) Risk Assessment and Early Warning

[0148] (3321)Risk Assessment

[0149] The deep learning model is used to comprehensively analyze the patient's physiological data to assess the risk level of heart failure. The risk level is usually divided into three levels: low, medium, and high, corresponding to different intervention measures.

[0150] (3322) Early Warning Mechanism

[0151] When the system detects early signs of heart failure, it automatically generates warning information and notifies patients and doctors through a mobile application. The warning information includes current physiological parameters, risk level and recommended interventions.

[0152] (3323)Intervention measures

[0153] Depending on the risk level, the system provides the following interventions:

[0154] (A1) Low risk: Patients are advised to increase the frequency of daily monitoring and maintain a healthy lifestyle.

[0155] (A2) Medium risk: Patients are advised to consult a doctor and may need to adjust medication doses or undergo further examinations.

[0156] (A3) High risk: Immediately notify the doctor and family members, and advise the patient to go to the hospital for emergency examination and treatment.

[0157] (333) Personalized rehabilitation advice and management methods

[0158] (3331) Health advice generation: The system generates personalized health management advice based on the patient's specific conditions and analysis results. Health advice includes:

[0159] (B1) Lifestyle adjustment: Recommendations include diet control, smoking cessation, alcohol restriction, and increasing appropriate physical exercise.

[0160] (B2) Medication use: Patients are advised to take medications prescribed by their doctor on time and monitor the effects of the medications regularly.

[0161] (B3) Regular check-ups: Patients are advised to undergo regular medical examinations such as electrocardiograms and blood tests to keep abreast of changes in their condition.

[0162] (3332) Rehabilitation plan formulation: Develop a detailed rehabilitation plan based on the model analysis results and the patient’s health status. The rehabilitation plan includes:

[0163] (C1) Dietary recommendations: A low-salt, low-fat, high-fiber diet is recommended to control body weight and maintain cardiovascular health.

[0164] (C2) Exercise plan: Develop an appropriate exercise plan based on the patient's physical condition, such as daily walking, swimming and other aerobic exercises.

[0165] (3333)Mobile Application Support

[0166] Through the mobile application, patients can receive and view personalized recovery suggestions in real time and communicate with doctors instantly. The application also provides reminders to help patients monitor and take medication on time.

[0167] (3334) Continuous Monitoring and Feedback

[0168] The system continuously monitors patients and regularly updates rehabilitation plans. Based on the latest data analysis and patient feedback, rehabilitation recommendations are adjusted to ensure the effectiveness and adaptability of rehabilitation programs. The system also provides a feedback mechanism, where patients can provide feedback on their health status and rehabilitation results through the app to help doctors optimize treatment plans.

Claims

1. A comprehensive non-invasive heart failure monitoring system, It is characterized in that It includes a multi-parameter monitoring wearable device, a cloud processor and an application layer. The monitoring data collected by the multi-parameter monitoring wearable device is uploaded to the cloud processor in real time through wireless communication. The cloud processor predicts and analyzes the collected monitoring data through machine learning algorithms. Patients and doctors view the monitoring data in real time through the application layer. The multi-parameter monitoring wearable device can continuously monitor heart rate, chest impedance, activity status and electrocardiogram; The cloud-based processor can predict the risk of heart failure and identify early signs of heart failure; The application layer can provide personalized health management suggestions and support instant communication between doctors and patients.

2. The all-round non-invasive heart failure monitoring system according to claim 1, It is characterized in that The multi-parameter monitoring wearable device includes a heart rate and electrocardiogram monitoring module, a respiration monitoring module, a chest impedance measurement module, an activity status monitoring module, a microcontroller, and a Wi-Fi module; The heart rate and electrocardiogram monitoring module uses photoplethysmography to measure blood flow changes and monitor heart rate by emitting and receiving light waves that penetrate the skin and blood vessels; The respiratory monitoring module detects inhalation or exhalation by measuring the chest pressure changes caused by respiratory movement, providing reliable data for respiratory monitoring; The thoracic impedance measurement module monitors potential fluid accumulation in heart failure patients by measuring impedance changes in the thoracic area; The activity status monitoring module monitors the patient's activity status by measuring the acceleration changes in three directions to identify the patient's resting, walking and other activity modes; The microcontroller collects the patient's physiological data in real time; All monitoring data is uploaded to the cloud database in real time via the built-in Wi-Fi module. The cloud processor conducts in-depth analysis of the collected data by identifying and analyzing heart failure models to predict heart failure events and assess heart failure risks.

3. The monitoring method of the all-round non-invasive heart failure monitoring system according to any one of claims 1 to 2, It is characterized in that The implementation steps are as follows: S1, wear the multi-parameter wearable device on the designated part to ensure good contact between the sensor and the skin; S2, after the device is started, the sensor collects physiological parameters in real time; the collected data is preliminarily processed by the device's built-in microcontroller and uploaded to the cloud database in real time stably and quickly through the optimized Wi-Fi module; S3, filtering, normalizing and feature extraction of the raw data uploaded to the cloud database; S4, extract key features from various physiological parameters through the extreme tree and perform preliminary classification; and input the features output by the extreme tree into the residual neural network; S5, using RNN and LSTM models to process time series data, capture physiological states that evolve over time, and identify early signs of heart failure; S6, comprehensively analyzing the patient's physiological data by identifying and analyzing the heart failure model to assess the risk level of heart failure; S7, when the monitoring system detects early signs of heart failure, it automatically generates warning information and notifies patients and doctors through mobile applications; and generates personalized health management suggestions based on the patient's specific condition and analysis results; S8, monitoring system continuously monitors the patient and regularly updates the rehabilitation plan; Adjust rehabilitation recommendations based on the latest data analysis and patient feedback to ensure the effectiveness and adaptability of the rehabilitation program.

4. The monitoring method of the all-round non-invasive heart failure monitoring system according to claim 3, It is characterized in that Heart rate is measured by photoplethysmography, ECG measures the heart's electrical activity through electrodes, chest impedance is measured by a tiny electrical current, and activity status is monitored by an accelerometer.

5. The monitoring method of the all-round non-invasive heart failure monitoring system according to claim 3, It is characterized in that The pre-processed data is presented in charts and numerical forms in a user-friendly mobile application, allowing patients and doctors to intuitively view real-time and historical data; Patients and doctors can understand their health status and view personalized health management recommendations through the user interface.

6. The monitoring method of the all-round non-invasive heart failure monitoring system according to claim 3, It is characterized in that In step S6, the monitoring system regularly updates the recognition and analysis model of heart failure to adapt to changes in the patient's condition and ensure the accuracy of the prediction; The implementation process of building a model for identifying and analyzing heart failure is as follows: S61, continuously collects patients’ physiological data through sensors integrated into multi-parameter wearable devices; S62, preprocessing the collected physiological data and extracting key features from the raw data; S63, using the ET-ResNet model to perform classification and regression analysis by evaluating the importance of individual physiological parameters and outputting prediction results; S64, regularly adjust the parameters of the ET-ResNet model based on the evaluation and feedback of the prediction results.

7. The monitoring method of the all-round non-invasive heart failure monitoring system according to claim 3, It is characterized in that The warning information includes current physiological parameters, risk level and intervention measures; according to the risk level, the monitoring system provides the following intervention measures: Low risk: Patients are advised to increase the frequency of daily monitoring and maintain a healthy lifestyle; Medium risk: Patients are advised to consult a doctor and adjust medication dosage or undergo further examination; High risk: Immediately notify doctors and family members and recommend patients to go to the hospital for emergency examination and treatment.

Citation Information

Cited By

  • Wearable chest positive resistance dynamic monitoring system and method for heart failure patient

    CN120514356A

  • Wearable chest positive airway pressure dynamic monitoring system for heart failure patients and method thereof

    CN120514356B

  • Heart failure compensation recognition system based on artificial intelligence

    CN121421551A

  • Artificial Intelligence-Based Heart Failure Compensation Identification System

    CN121421551B