Internal fistula monitoring health management bracelet and system

Abstract

The invention discloses an internal fistula monitoring health management bracelet and system, and relates to the technical field of medical health monitoring. Real-time and accurate monitoring and classification of the internal fistula health state are achieved, the problems that a traditional monitoring method cannot achieve real-time continuous monitoring and signals are prone to being interfered are solved, the multi-modal physiological signals are preprocessed through wavelet transformation and a signal enhancement algorithm, features are extracted in combination with the convolutional neural network and the long-short term memory network, and the real-time monitoring of the internal fistula health state is achieved. The monitoring accuracy and reliability are obviously improved; meanwhile, a personalized early warning threshold value is calculated by combining a deep learning model with individual difference data of the patient, the internal fistula state is accurately judged, early warning is given out in time, the problem that the internal fistula state is difficult to comprehensively reflect by a single signal is solved, and it is ensured that the patient can obtain medical intervention in time.

Description

Technical Field

[0001] The present invention relates to the technical field of medical health monitoring, and specifically, to an internal fistula monitoring health management bracelet and system. Background Art

[0002] The internal fistula of the vascular access is one of the common complications of dialysis patients. The assessment of its health status is crucial for the quality of life of patients. Currently, the monitoring of the internal fistula mainly relies on regular ultrasound examinations, but continuous real-time monitoring cannot be achieved; moreover, physiological signals such as tremors, blood flow sounds, blood pressure, pulse, and respiration in the internal fistula area are closely related to its health status. However, these signals are weak and vulnerable to environmental noise interference, making it difficult to capture and analyze; at the same time, there are individual differences in the physiological signals of different patients, and a single signal is difficult to comprehensively reflect the state of the internal fistula. Summary of the Invention

[0003] The purpose of the present invention is to provide an internal fistula monitoring health management bracelet and system, which realizes real-time monitoring, accurate classification, personalized early warning, and data management of the health status of the internal fistula, provides comprehensive health monitoring and timely medical intervention support for patients, and significantly improves the intelligent level of internal fistula health management and the safety of patients.

[0004] The purpose of the present invention can be achieved through the following technical solutions: The present application provides an internal fistula monitoring health management system, which processes and analyzes by receiving the data of the internal fistula monitoring health management bracelet, including: A data acquisition and processing module, which acquires physiological signals such as tremors, blood flow sounds, blood pressure, pulse, and respiration in the internal fistula area, preprocesses the physiological signals, filters out environmental noise, enhances the signal intensity, and obtains a denoised multi-modal physiological signal data set; A feature extraction module, which, according to the multi-modal physiological signal data set, uses a convolutional neural network to extract features from the signals, obtains a set of high-dimensional feature vectors, and then establishes a time series model through a long short-term memory network to capture the dynamic change trend of the signals in the time dimension and obtains time series features; A fusion and classification module, which fuses the set of high-dimensional feature vectors and the time series features, constructs a multi-dimensional feature space, and uses a support vector machine to classify the feature space to obtain the classification result of the internal fistula health status; An intelligent early warning module, when the classification result of the internal fistula health status shows an abnormality, triggers an early warning mechanism, generates an abnormal early warning signal and sends it to the monitoring terminal, and according to the abnormal early warning signal, combines the individual difference data of the patient, and calculates a personalized early warning threshold through a deep learning model to determine the final early warning level; A display and storage module, which displays the classification result of the internal fistula health status and the early warning level on the monitoring terminal, and at the same time stores the data processing process and results in the database.

[0005] Further, the data acquisition and processing module includes: Obtain the original data of the tremor signal, blood flow sound signal, blood pressure signal, pulse signal, and respiration signal in the arteriovenous fistula area through sensors, perform downsampling processing on the original data, and unify the sampling rate to the standard frequency; Decompose the downsampled signal using wavelet transform, extract the high-frequency component and low-frequency component, set a noise threshold for the high-frequency component, filter out the high-frequency components greater than the preset threshold, and retain the effective signal components; Then use the signal enhancement algorithm to estimate the power spectral density of the filtered low-frequency component, enhance the signal intensity within a specific frequency band range, suppress the signal energy in the non-target frequency band, and improve the signal quality; Classify the enhanced signals according to the signal type, and establish a tremor signal feature set, a blood flow sound signal feature set, a blood pressure signal feature set, a pulse signal feature set, and a respiration signal feature set respectively.

[0006] Further, after classifying the enhanced signals, it also includes: extracting features from the classified signal feature sets, selecting time-domain features, frequency-domain features, and non-linear features, constructing a multi-dimensional feature space, obtaining the time-frequency feature matrix of various signals, training a multi-modal signal classification model based on the support vector machine algorithm, setting training samples of different signal types, optimizing the classifier parameters, and obtaining a signal type discrimination function; then using the semi-supervised learning method to automatically annotate the multi-modal signal data set, combining the annotated samples and unannotated samples, iteratively optimizing the annotation results, and generating a standardized multi-modal physiological signal database.

[0007] Further, the feature extraction module includes: Obtain the original signal in the multi-modal physiological signal data set and input it into the convolutional neural network model for convolution operation, then extract the output features of each convolutional layer in the convolutional neural network, and transform the feature map into a high-dimensional feature vector; Arrange the high-dimensional feature vectors in chronological order to construct time series feature data, and use the long short-term memory network to model the time series feature data to capture the dynamic change law in the time dimension; Extract the output state of the hidden layer in the long short-term memory network to obtain the change trend feature of the signal in the time dimension.

[0008] Further, the fusion and classification module includes: Fuse the high-dimensional feature vector set and the time series feature, fuse the high-dimensional feature vector set and the time series feature, construct a multi-dimensional feature space, and use the support vector machine to classify the multi-dimensional feature space. When the accuracy of the arteriovenous fistula health state in the classification result reaches the preset threshold, output the final classification result of the arteriovenous fistula health state; When the classification result does not reach the preset threshold, the hyperparameters of the convolutional neural network and the long short-term memory network model are readjusted to optimize the feature extraction and time series modeling process; Obtain the optimized high-dimensional feature vector set and time series features, re-perform feature fusion and construction of the multi-dimensional feature space, use a support vector machine to classify again, judge the health status of the internal fistula according to the re-classification result, and output the final classification result of the internal fistula health status.

[0009] Further, the intelligent warning module includes: Obtain the classification result of the internal fistula health status. When the classification result shows an outlier, trigger the warning mechanism to generate an abnormal warning signal, send the abnormal warning signal to the monitoring terminal, and at the same time extract the individual data of the patient from the database; Use deep learning to extract features from the individual data, calculate the personalized threshold limit, and judge whether the warning level meets the preset conditions according to the comparison result between the personalized threshold limit and the outlier; Input the judgment result into the warning mechanism, generate the final warning level and update the display content of the monitoring terminal. When the warning level reaches the highest level, start the emergency processing module and send an emergency notice to the relevant medical personnel.

[0010] Further, the display and storage module includes: Display the classification result of the internal fistula health status and the warning level on the monitoring terminal. At the same time, obtain the internal fistula health status data stream through the monitoring terminal, analyze and process the data stream according to the pre-established health status classification model to obtain the health status classification result, and then match the threshold range corresponding to the warning level according to the classification result to judge the warning level of the internal fistula health status; According to the warning level, use the processing chain to encapsulate the classification result and the warning level into structured data, write the structured data into the repository and generate a data index; and perform a correlation analysis on the historical data in the repository, update the parameters of the health status classification model, and synchronize the updated parameters to the classification model through the logic chain to form a closed-loop processing flow.

[0011] An internal fistula monitoring health management bracelet, applied to an internal fistula monitoring health management system, includes: A health status display unit for displaying the health status of the internal fistula of the wearer, including icons or text information of the normal state, the state of internal fistula blood vessel occlusion, and the state information of signs of problems with the internal fistula; A wireless data transmission unit that transmits monitoring data to a smart phone or a remote medical center in real time via Bluetooth or Wi-Fi; A storage module for storing monitored respiratory data, internal fistula flutter data, heart rate data, and blood pressure data; A positioning module, which is used to send location information to the control module. In case of an emergency, it is convenient for medical staff to quickly locate the patient's position. An alarm module, which emits audible, visual, and electrical signals to alert the patient or medical staff when the monitored data exceeds the set threshold, so as to promptly detect abnormal situations. A monitoring module, which is used to collect and process the physiological signals of the wearer in real time and evaluate the health status of the arteriovenous fistula. A control module, which is used to receive the data of each sensor, analyze and process it, and determine whether to trigger an alarm. Furthermore, the monitoring module includes: a vibration detection module, which is used to detect the vibration characteristics of the arteriovenous fistula area through a highly sensitive piezoelectric film sensor. It is set on the inner side of the bracelet, closely attached to the wearer's skin and facing the area where the arteriovenous fistula is located. By measuring the tremor signals on the wearer's skin, including the intensity and frequency of the tremor signals in the three-axis directions, the blood flow condition of the arteriovenous fistula is evaluated. A blood vessel monitoring module, which is used to non-invasively obtain the measurement of the blood vessel wall thickness and blood flow velocity. It obtains the flutter sound data of the arteriovenous fistula through a blood flow sound sensor. A highly sensitive stethoscope probe is attached to the arteriovenous fistula area to capture the blood flow sound signal at the arteriovenous fistula, including the frequency, amplitude, pitch, and volume parameters of the flutter sound. By amplifying, filtering, and analyzing the sound signal, it is judged whether the blood flow state of the arteriovenous fistula is normal. A blood pressure monitoring module, which is used to monitor the patient's blood pressure data. It real-time detects the patient's systolic and diastolic blood pressures through a blood pressure sensor, filters, amplifies, and performs analog-to-digital conversion on the monitored blood pressure data, and then transmits it to the control module. When the blood pressure data is lower or higher than the set threshold, the control module will trigger the alarm module to issue an alarm to remind the patient to seek medical treatment in time. A pulse monitoring module, which monitors the patient's heart rate data through a pulse sensor. The pulse signal captured by the pulse sensor will be amplified and filtered by a processor, and then the heart rate data is calculated by a heart rate calculation module. When the heart rate is lower or higher than the set threshold, the control module will issue an alarm signal. A respiration monitoring module, which monitors the patient's respiration data using an infrared sensor. It records the respiration frequency and depth by detecting the change in the thermal airflow during the patient's respiration. The monitored respiration data is amplified and filtered, and then transmitted to the control module. When the respiration data is lower or higher than the set threshold, the control module will trigger the alarm module to issue an alarm. It also includes a temperature sensor and pressure monitoring. The temperature sensor is used to monitor the skin temperature at the arteriovenous fistula to assist in judging the state of the arteriovenous fistula. The pressure monitoring monitors the pressure change in the arteriovenous fistula area through a pressure sensor, and judges whether there is stenosis or other abnormal conditions in the arteriovenous fistula by analyzing the pressure data.

[0012] The beneficial effects of the present invention are: The health management bracelet for fistula monitoring can collect multi-modal physiological signals such as tremors, blood flow sounds, blood pressure, pulse, and respiration in the fistula area in real time. It uses wavelet transform and signal enhancement algorithms to perform downsampling, denoising, and enhancement on the collected signals, and then uses convolutional neural networks and long short-term memory networks to extract features and perform time series modeling on the signals. Finally, it realizes the accurate classification of the health status of the fistula, solves the problems that traditional methods cannot monitor in real time and signals are difficult to capture and analyze, significantly improves the accuracy of monitoring, and can accurately distinguish the normal state, the state of fistula vessel occlusion, and the state of signs of problems in the fistula; By using a deep learning model and combining the individual difference data of patients (such as age, medical history, etc.) to calculate personalized warning thresholds, when the monitoring data exceeds the set threshold, the bracelet will emit audio-visual signals to remind the patient or medical staff. And when the warning level reaches the highest level, it will automatically activate the emergency processing module and send an emergency notice to relevant medical personnel, improving the accuracy and pertinence of the warning, solving the problem that a single signal is difficult to comprehensively reflect the state of the fistula, and ensuring that patients can obtain medical intervention in time; Through the display and storage module, the classification results of the fistula health status and the warning level are displayed on the monitoring terminal in real time, and the data processing process and results are stored in the database. Through the correlation analysis of historical data, the parameters of the health status classification model are dynamically updated to form a closed-loop processing flow, improving the efficiency of data management and the adaptability of the model, further optimizing the model performance, and providing an efficient data management and continuously optimized solution for fistula health management. Brief Description of the Drawings

[0013] For better understanding and implementation, the technical solutions of this application will be described in detail below with reference to the accompanying drawings.

[0014] Figure 1 It is a schematic structural diagram of a fistula monitoring health management system provided in Embodiment 1 of this application; Figure 2 It is a schematic flow diagram of the data acquisition and processing module of a fistula monitoring health management system provided in Embodiment 1 of this application; Figure 3 It is a schematic structural diagram of a fistula monitoring health management bracelet provided in Embodiment 2 of this application; Detailed Implementation Modes

[0015] To further illustrate the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the exemplary embodiments will be described in detail herein, and the examples are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of methods and systems consistent with some aspects of the present application as detailed in the appended claims.

[0016] The terms used in the present application are for the purpose of describing specific embodiments only and are not intended to limit the present application. The singular forms "a", "the", and "said" used in the present application and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0017] The following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation manners, features, and effects of the present invention.

[0018] Embodiment 1 Please refer to Figure 1 - Figure 2 , this embodiment provides an internal fistula monitoring health management system, which processes and analyzes by receiving the data of the internal fistula monitoring health management bracelet, including: A data acquisition and processing module, which acquires the tremor, blood flow sound, blood pressure, pulse, and respiratory physiological signals in the internal fistula area, preprocesses the physiological signals, filters out environmental noise, enhances the signal intensity, and obtains a denoised multi-modal physiological signal data set; Further, the data acquisition and processing module includes: S11. Obtain the original data of the tremor signal, blood flow sound signal, blood pressure signal, pulse signal, and respiratory signal in the internal fistula area through sensors, perform downsampling processing on the original data, and unify the sampling rate to the standard frequency; S12. Decompose the downsampled signal using wavelet transform, extract the high-frequency component and the low-frequency component, set a noise threshold for the high-frequency component, filter out the high-frequency components greater than the preset threshold, and retain the effective signal components; S13. Then use a signal enhancement algorithm to estimate the power spectral density of the filtered low-frequency component, enhance the signal intensity within a specific frequency band range, suppress the signal energy in the non-target frequency band, and improve the signal quality; S14. Classify the enhanced signals according to the signal type, and respectively establish a tremor signal feature set, a blood flow sound signal feature set, a blood pressure signal feature set, a pulse signal feature set, and a respiratory signal feature set.

[0019] Further, after classifying the enhanced signals, it also includes: extracting features from the classified signal feature set, selecting time-domain features, frequency-domain features, and non-linear features, constructing a multi-dimensional feature space, obtaining the time-frequency feature matrix of each type of signal, training a multi-modal signal classification model based on the support vector machine algorithm, setting training samples of different signal types, optimizing the classifier parameters, and obtaining a signal type discrimination function; then using a semi-supervised learning method to automatically label the multi-modal signal data set, combining the labeled samples and unlabeled samples, iteratively optimizing the labeling results, and generating a standardized multi-modal physiological signal database.

[0020] It should be explained that: in the signal processing and classification process of the fistula monitoring health management bracelet system, the "signal type discrimination function" is a key function based on the support vector machine (SVM) algorithm for identifying and classifying different signal types. By mapping the input multi-modal physiological signal feature vector into a high-dimensional feature space, it uses the classification hyperplane trained by the support vector machine (SVM) algorithm for judgment. Specifically, it calculates the distance between the input feature vector and the classification hyperplane, and judges the category of the signal according to the positive or negative of the distance. If the distance is positive, the signal is classified into one category (such as the "normal" state); if the distance is negative, the signal is classified into another category (such as the "abnormal" state); this distance-based classification method can effectively distinguish different types of signals, so as to realize the rapid and accurate judgment of the fistula health status.

[0021] Specifically, through the processing of collecting, downsampling, wavelet decomposition, noise filtering, and signal enhancement of the multi-modal physiological signals in the fistula area, the signal quality is effectively improved. At the same time, combined with feature extraction, classification modeling, and semi-supervised learning, a standardized multi-modal physiological signal database is constructed, significantly improving the signal analyzability and diagnostic value.

[0022] The feature extraction module uses a convolutional neural network to extract features from the multi-modal physiological signal data set to obtain a set of high-dimensional feature vectors, and then establishes a time series model through a long short-term memory network to capture the dynamic change trend of the signal in the time dimension and obtain time series features; Further, the feature extraction module includes: Obtain the original signal in the multi-modal physiological signal data set, input it into the convolutional neural network model for convolutional operation, then extract the output features of each convolutional layer in the convolutional neural network, and convert the feature map into a high-dimensional feature vector; Arrange the high-dimensional feature vectors in chronological order to construct time series feature data, and use a long short-term memory network to model the time series feature data to capture the dynamic change law in the time dimension; Extract the output state of the hidden layer in the long short-term memory network to obtain the change trend characteristics of the signal in the time dimension.

[0023] Specifically, the convolutional neural network is used to perform deep feature extraction on the multimodal physiological signals to generate a set of high-dimensional feature vectors. Then, the long short-term memory network is used to model these features in a time series to accurately capture the dynamic change trend of the signal in the time dimension. Finally, the time series features reflecting the dynamic characteristics of the signal are extracted, providing richer and more representative feature information for signal analysis and diagnosis, and improving the depth and accuracy of signal processing.

[0024] The fusion and classification module fuses the high-dimensional feature vector set and the time series features to construct a multi-dimensional feature space, and uses a support vector machine to classify the feature space to obtain the classification result of the health status of the internal fistula; Further, the fusion and classification module includes: Fuse the high-dimensional feature vector set and the time series features, fuse the high-dimensional feature vector set and the time series features to construct a multi-dimensional feature space, and use a support vector machine to classify the multi-dimensional feature space. When the accuracy rate of the health status of the internal fistula in the classification result reaches the preset threshold, the final classification result of the health status of the internal fistula is output; When the classification result does not reach the preset threshold, the hyperparameters of the convolutional neural network and the long short-term memory network model are readjusted to optimize the feature extraction and time series modeling process; Obtain the optimized high-dimensional feature vector set and time series features, reconstruct the feature fusion and the multi-dimensional feature space, use a support vector machine to classify again, judge the health status of the internal fistula according to the re-classification result, and output the final classification result of the health status of the internal fistula.

[0025] Among them, fusing the high-dimensional feature vector and the time series features is to directly splice these features into a larger feature vector to form a multi-dimensional feature space; for example, if the dimension of the high-dimensional feature vector is d1 and the dimension of the time series feature is d2, the dimension of the fused feature vector is d1 + d2. Then, through a feature fusion architecture such as BiFPN (Bidirectional Feature Pyramid Network) to enhance the information flow between features and improve the interoperability of different-level features. BiFPN can effectively integrate multi-scale features through a bidirectional feature pyramid network structure, further enhancing the model's ability to capture complex patterns.

[0026] Specifically, by fusing high-dimensional feature vectors with time-series features to construct a multi-dimensional feature space and using a support vector machine for classification, an accurate judgment of the health status of the internal fistula is achieved. When the classification accuracy does not reach the threshold, the hyperparameters of the convolutional neural network and the long short-term memory network are adjusted to optimize feature extraction, and the BiFPN architecture is combined to enhance the flow and integration of feature information, further improving the model's ability to capture complex patterns, and finally outputting a reliable classification result of the internal fistula health status, significantly improving the accuracy and reliability of internal fistula health monitoring.

[0027] Intelligent warning module. When the classification result of the internal fistula health status shows an abnormality, it triggers a warning mechanism, generates an abnormal warning signal and sends it to the monitoring terminal. According to the abnormal warning signal, combined with the individual difference data of the patient, a personalized warning threshold is calculated through a deep learning model to determine the final warning level. Furthermore, the intelligent warning module includes: Obtain the classification result of the internal fistula health status. When the classification result shows an abnormal value, trigger a warning mechanism to generate an abnormal warning signal, send the abnormal warning signal to the monitoring terminal, and at the same time extract the individual data of the patient from the database. Use deep learning to extract features from individual data, calculate personalized threshold limits, and judge whether the warning level meets the preset conditions according to the comparison result between the personalized threshold limits and the abnormal values. Input the judgment result into the warning mechanism, generate the final warning level and update the display content of the monitoring terminal. When the warning level reaches the highest level, start the emergency treatment module and send an emergency notice to relevant medical personnel.

[0028] Among them, a personalized warning threshold is calculated through a deep learning model combined with the individual difference data of the patient (such as age, medical history, basic health status, etc.) to improve the accuracy and pertinence of the warning. When the warning level reaches the highest level, start the emergency treatment module and send an emergency notice to relevant medical personnel to ensure that the patient can obtain medical intervention in a timely manner.

[0029] Specifically, by accurately monitoring the health status of the internal fistula, when an abnormality is detected, the warning mechanism is quickly triggered. Combined with the individual difference data of the patient (such as age, medical history, etc.), a personalized warning threshold is calculated using a deep learning model, so as to accurately determine the warning level and update the display content of the monitoring terminal in real time. When the warning level reaches the highest level, the emergency treatment module is automatically started and an emergency notice is sent to medical personnel to ensure that the patient can obtain targeted medical intervention in a timely manner, significantly improving the intelligent level and emergency response ability of internal fistula health monitoring and ensuring patient safety.

[0030] Display and storage module. Display the classification result of the internal fistula health status and the warning level on the monitoring terminal, and at the same time store the data processing process and results in the database.

[0031] Further, the display and storage module includes: The classification result of the internal fistula health status and the warning level are displayed on the monitoring terminal. At the same time, the data stream of the internal fistula health status is obtained through the monitoring terminal, analyzed and processed according to the pre-established health status classification model to obtain the classification result of the health status, and then the threshold range corresponding to the warning level is matched according to the classification result to judge the warning level of the internal fistula health status; According to the warning level, the classification result and the warning level are encapsulated with a processing chain to form structured data, which is written into the repository and a data index is generated; the historical data in the repository is analyzed for relevance, the parameters of the health state classification model are updated, and the updated parameters are synchronized to the classification model through a logic chain to form a closed-loop processing flow.

[0032] Specifically, the real-time visualization display of the classification result of the internal fistula health status and the warning level is realized, and the data stream of the health status is dynamically analyzed and processed through the monitoring terminal to match the warning level. At the same time, the processing result is encapsulated as structured data and stored in the database, and a data index is generated for easy query and traceability. In addition, through the relevance analysis of historical data, the parameters of the health status classification model are dynamically updated, and the closed-loop synchronization of the parameters is realized, further optimizing the model performance, ensuring the accuracy and reliability of data processing, and providing an efficient data management and continuous optimization solution for internal fistula health monitoring.

[0033] Embodiment 2 Please refer to Figure 3 , this embodiment provides an internal fistula monitoring health management bracelet, which applies an internal fistula monitoring health management system, including: A health status display unit for displaying the health status of the internal fistula of the wearer, including icons or text information of the normal state, the state of internal fistula blood vessel occlusion, and the state information of signs of problems with the internal fistula, facilitating patients and medical staff to quickly understand the health status of the internal fistula.

[0034] A wireless data transmission unit that transmits the monitoring data to a smart phone or a remote medical center in real time via Bluetooth or Wi-Fi, enabling medical staff to remotely view the health status of the patient's internal fistula and realizing remote medical treatment and real-time management; A storage module for storing the monitored respiratory data, internal fistula tremor data, heart rate data, and blood pressure data, facilitating subsequent data analysis and historical record query; A positioning module for sending location information to the control module, which is convenient for medical staff to quickly locate the patient's location in case of emergency; An alarm module that emits audible, light, and electrical signals to alert the patient or medical staff when the monitoring data exceeds the set threshold, promptly detecting abnormal situations and avoiding missing the best intervention time; A monitoring module for collecting and processing the physiological signals of the wearer in real time to evaluate the health status of the arteriovenous fistula; A control module for receiving the data from each sensor, analyzing and processing it, and determining whether to trigger an alarm; Further, the monitoring module includes: a vibration detection module for detecting the vibration characteristics of the arteriovenous fistula region through a highly sensitive piezoelectric film sensor, which is arranged on the inner side of the bracelet, closely attached to the skin of the wearer and facing the region where the arteriovenous fistula is located. By measuring the tremor signals on the skin of the wearer, including the intensity and frequency of the tremor signals in the three-axis directions, the blood flow condition of the arteriovenous fistula is evaluated; A blood vessel monitoring module for non-invasively obtaining the measurement of the blood vessel wall thickness and blood flow velocity, and obtaining the flutter sound data of the arteriovenous fistula through a blood flow sound sensor. It is through a highly sensitive stethoscope probe attached to the arteriovenous fistula region to capture the blood flow sound signals at the arteriovenous fistula. These signals include parameters such as the frequency, amplitude, pitch, and volume of the flutter sound. By amplifying, filtering, and analyzing these sound signals through a signal processing unit, it can be judged whether the blood flow state of the arteriovenous fistula is normal; A blood pressure monitoring module for monitoring the blood pressure data of the patient. Its core component is a blood pressure sensor, which can monitor the systolic blood pressure and diastolic blood pressure of the patient in real time. The monitored blood pressure data is filtered, amplified, and analog-to-digital converted, and then transmitted to the control module. When the blood pressure data is lower or higher than the set threshold, the control module will trigger the alarm module to issue an alarm to remind the patient to seek medical treatment in time; A pulse monitoring module for monitoring the heart rate data of the patient through a pulse sensor. The pulse signals captured by the pulse sensor will be amplified and filtered by a processor, and then the heart rate data is calculated by a heart rate calculation module. If the heart rate is lower or higher than the set threshold, the control module will issue an alarm signal. This monitoring method can timely detect the abnormal heart rate of the patient and help evaluate the overall health status of the patient; A respiration monitoring module for monitoring the respiration data of the patient by using an infrared sensor. The infrared sensor records the respiration frequency and depth by detecting the change of the thermal airflow when the patient breathes. The monitored respiration data is amplified and filtered, and then transmitted to the control module. When the respiration data is lower or higher than the set threshold, the control module will trigger the alarm module to issue an alarm; Among them, the monitoring module also includes: a temperature sensor and pressure monitoring. The temperature sensor is used to monitor the skin temperature at the arteriovenous fistula to assist in judging the state of the arteriovenous fistula; the pressure monitoring monitors the pressure change in the arteriovenous fistula region through a pressure sensor, and judges whether there is stenosis or other abnormal conditions in the arteriovenous fistula by analyzing the pressure data.

[0035] Specifically, in the fistula monitoring health management bracelet, each sensor module collects the patient's physiological signals in real time, such as tremor signals, blood flow sound signals, blood pressure signals, pulse signals, respiratory signals, body temperature signals, and pressure signals. These signals are first preprocessed, including downsampling and filtering, to remove environmental noise and unify the sampling rate. Subsequently, the signals are further decomposed by wavelet transform to extract high-frequency and low-frequency components and filter out high-frequency noise. Then, a signal enhancement algorithm is used to estimate the power spectral density of the low-frequency components to enhance the signal intensity. Finally, these processed signals are classified and feature extracted to form a multi-modal physiological signal dataset, which serves as the input data for the data acquisition and processing module.

[0036] The bracelet visually displays the fistula health status through the health status display unit and uses the wireless data transmission unit to transmit the data to the smartphone or the remote medical center in real time, facilitating remote monitoring by medical staff. At the same time, the bracelet is equipped with an alarm module to timely remind the patient and medical staff when the monitoring data is abnormal, ensuring that the patient can receive timely intervention in case of emergency. In addition, the bracelet also has a positioning function, which is convenient for quickly locating the patient's position in case of emergency, thus realizing real-time monitoring, intelligent early warning, and remote management of the fistula health status, effectively improving the patient's health management level and quality of life.

[0037] The above is only a preferred embodiment of the present invention and does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the equivalent embodiments with equivalent changes within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A fistula monitoring health management system, characterized in that: The data received from the fistula monitoring health management bracelet is processed and analyzed, including: The data acquisition and processing module obtains the tremor, blood flow sound, blood pressure, pulse and respiratory physiological signals in the fistula area, pre-processes the physiological signals, filters out environmental noise, enhances signal strength, and obtains a denoised multimodal physiological signal data set; The feature extraction module uses a convolutional neural network to extract features from multimodal physiological signal data sets to obtain a high-dimensional feature vector set. It then uses a long short-term memory network to build a time series model to capture the dynamic change trend of the signal in the time dimension and obtain the time series features. The fusion and classification module fuses the high-dimensional feature vector set with the time series features to construct a multi-dimensional feature space, and uses a support vector machine to classify the feature space to obtain the classification results of the health status of the fistula; Intelligent early warning module: When the classification result of the health status of the fistula shows abnormality, the early warning mechanism is triggered, an abnormal early warning signal is generated and sent to the monitoring terminal. Based on the abnormal early warning signal and the individual difference data of the patient, the personalized early warning threshold is calculated through the deep learning model to determine the final early warning level; The display and storage module displays the classification results and warning levels of the fistula health status on the monitoring terminal, and stores the data processing process and results in the database.

2. The fistula monitoring health management system according to claim 1, characterized in that: The data acquisition and processing module comprises: The original data of tremor signal, blood flow sound signal, blood pressure signal, pulse signal and respiratory signal in the fistula area are obtained through sensors, and the original data are downsampled to unify the sampling rate to the standard frequency; The downsampled signal is decomposed by wavelet transform, high-frequency components and low-frequency components are extracted, a noise threshold is set for the high-frequency components, high-frequency components greater than the preset threshold are filtered out, and effective signal components are retained; Then, the signal enhancement algorithm is used to estimate the power spectrum density of the filtered low-frequency component, enhance the signal strength within a specific frequency band, and suppress the signal energy of non-target frequency bands; The enhanced signals are classified according to the signal types, and the vibration signal feature set, blood flow sound signal feature set, blood pressure signal feature set, pulse signal feature set and breathing signal feature set are established respectively.

3. The internal fistula monitoring health management system according to claim 2, characterized in that: After classifying the enhanced signal, it also includes: extracting features from the classified signal feature set, selecting time domain features, frequency domain features and nonlinear features, constructing a multidimensional feature space, obtaining time-frequency feature matrices of various signals, training a multimodal signal classification model based on a support vector machine algorithm, setting training samples of different signal types, optimizing classifier parameters, and obtaining a signal type discriminant function; and then using a semi-supervised learning method to automatically annotate the multimodal signal data set, combining annotated samples and unlabeled samples, iteratively optimizing the annotation results, and generating a standardized multimodal physiological signal database.

4. The fistula monitoring health management system according to claim 1, characterized in that: The feature extraction module comprises: The original signal in the multimodal physiological signal data set is obtained and input into the convolutional neural network model for convolution operation, and then the output features of each convolutional layer in the convolutional neural network are extracted to convert the feature map into a high-dimensional feature vector; Arrange high-dimensional feature vectors in chronological order to construct time series feature data, and use long short-term memory networks to model the time series feature data to capture the dynamic changes in the time dimension; Extract the output state of the hidden layer in the long short-term memory network to obtain the changing trend characteristics of the signal in the time dimension.

5. The fistula monitoring health management system according to claim 1, characterized in that: The fusion and classification module includes: The high-dimensional feature vector set is fused with the time series feature, and a multi-dimensional feature space is constructed. The multi-dimensional feature space is classified using a support vector machine. When the accuracy of the fistula health status in the classification result reaches a preset threshold, the final fistula health status classification result is output; When the classification result does not reach the preset threshold, the hyperparameters of the convolutional neural network and long short-term memory network models are readjusted to optimize the feature extraction and time series modeling process; Obtain the optimized high-dimensional feature vector set and time series features, re-perform feature fusion and multi-dimensional feature space construction, use support vector machine to perform classification again, judge the health status of the fistula based on the reclassification results, and output the final classification result of the health status of the fistula.

6. The fistula monitoring health management system according to claim 1, characterized in that: The intelligent early warning module comprises: Obtain the classification results of the health status of the fistula. When the classification results show abnormal values, the early warning mechanism is triggered to generate abnormal early warning signals, which are sent to the monitoring terminal, and the individual data of the patient is extracted from the database; Use deep learning to extract features from individual data, calculate personalized thresholds, and determine whether the warning level meets the preset conditions based on the comparison between the personalized thresholds and abnormal values; The judgment results are input into the early warning mechanism to generate the final warning level and update the display content of the monitoring terminal. When the warning level reaches the highest level, the emergency processing module is activated to send an emergency notification to relevant medical personnel.

7. The fistula monitoring health management system according to claim 1, characterized in that: The display and storage module includes: The monitoring terminal displays the classification results and warning levels of the internal fistula health status. At the same time, the monitoring terminal obtains the internal fistula health status data stream, analyzes and processes the data stream according to the pre-established health status classification model, obtains the health status classification results, and then matches the threshold range corresponding to the warning level according to the classification results to determine the warning level of the internal fistula health status; According to the warning level, the processing chain is used to encapsulate the classification results and warning levels to form structured data, write the structured data into the repository and generate a data index; perform correlation analysis on the historical data in the repository, update the parameters of the health classification model, and synchronize the updated parameters to the classification model through the logic chain to form a closed-loop processing flow.

8. A fistula monitoring health management bracelet, using a fistula monitoring health management system as claimed in any one of claims 1 to 7, characterized in that: include: A health status display unit, used to display the wearer's fistula health status, including icons or text information of normal status, fistula vascular occlusion status, and fistula problem signs; Wireless data transmission unit, which transmits monitoring data to a smartphone or remote medical center in real time via Bluetooth or Wi-Fi; A storage module, used for storing monitored respiratory data, fistula tremor data, heart rate data and blood pressure data; The positioning module is used to send location information to the control module, so that medical staff can quickly locate the patient in an emergency; The alarm module sends out sound and light signals to remind patients or medical staff when the monitoring data exceeds the set threshold, so as to detect abnormal conditions in time; The monitoring module is used to collect and process the wearer's physiological signals in real time to evaluate the health status of the fistula; The control module is used to receive data from each sensor, analyze and process it, and determine whether to trigger an alarm.

9. The fistula monitoring health management bracelet according to claim 8, characterized in that: The monitoring module includes: a vibration detection module, which is used to detect the vibration characteristics of the fistula area through a high-sensitivity piezoelectric film sensor, which is arranged on the inner side of the bracelet, closely fits the wearer's skin and faces the fistula area, and evaluates the blood flow of the fistula by measuring the tremor signal on the wearer's skin, including the intensity and frequency of the tremor signal in three-axis directions; The vascular monitoring module is used to non-invasively obtain and measure the thickness of the vascular wall and the blood flow velocity. It obtains the fistula trill data through the blood flow sound sensor. The high-sensitivity stethoscope probe is attached to the fistula area to capture the blood flow sound signal at the fistula, including the frequency, amplitude, pitch and volume parameters of the trill. The sound signal is amplified, filtered and analyzed to determine whether the blood flow status of the fistula is normal. The blood pressure monitoring module is used to monitor the patient's blood pressure data. The blood pressure sensor detects the patient's systolic and diastolic blood pressure in real time, filters, amplifies and converts the monitored blood pressure data into analog-to-digital data, and then transmits it to the control module. When the blood pressure data is lower than or higher than the set threshold, the control module will trigger the alarm module to sound an alarm, reminding the patient to seek medical treatment in time; The pulse monitoring module monitors the patient's heart rate data through a pulse sensor. The pulse signal captured by the pulse sensor will be amplified and filtered by the processor, and then the heart rate data will be calculated by the heart rate calculation module. When the heart rate is lower or higher than the set threshold, the control module will send out an alarm signal; The respiratory monitoring module uses an infrared sensor to monitor the patient's respiratory data, records the respiratory rate and depth by detecting the changes in thermal airflow when the patient breathes, amplifies and filters the monitored respiratory data, and then transmits it to the control module. When the respiratory data is lower than or higher than the set threshold, the control module will trigger the alarm module to sound an alarm.

10. The fistula monitoring health management bracelet according to claim 8, characterized in that: Also includes: Temperature sensor and pressure monitoring: The temperature sensor is used to monitor the skin temperature at the arteriovenous fistula to assist in determining the state of the fistula. Pressure monitoring uses a pressure sensor to monitor pressure changes in the fistula area and analyzes pressure data to determine whether the fistula has stenosis or other abnormalities.

Images