Portable heart rate abnormity early warning system based on edge calculation

Abstract

The invention relates to the technical field of heart rate monitoring, in particular to a portable heart rate anomaly early warning system based on edge computing, which comprises a signal acquisition module, a signal preprocessing module, a data fusion module, an anomaly detection module, a local alarm module and a cloud interaction module. According to the system, physiological signals are collected through a micro sensor array, filtering, denoising and feature extraction are carried out through edge computing nodes, comprehensive heart rate data are generated through a multi-modal data fusion algorithm, and the heart rate state is judged in combination with an anomaly detection model. And if abnormity is detected, the local alarm module gives out early warning and uploads the data to the cloud for further analysis. Real-time monitoring and low-delay processing are achieved, the accuracy and reliability of heart rate monitoring are improved, and accurate health early warning service is provided for a user.

Description

Technical Field

[0001] The present invention belongs to the cross - technical field of medical health and information technology, and specifically relates to a portable heart rate anomaly warning system based on edge computing. Background Art

[0002] In application scenarios of heart rate monitoring and anomaly warning, such as personal health management, sports monitoring, or medical monitoring fields, in order to improve the monitoring efficiency and ensure the health and safety of users, it is necessary to collect and analyze heart rate data in real - time and accurately, so as to timely detect abnormal situations and issue warnings. However, traditional monitoring methods usually rely on centralized data processing methods, which may be affected by network latency or bandwidth limitations during data transmission, resulting in a reduction in the real - time performance of monitoring results. In addition, due to the combined effects of various factors such as individual differences, motion interference, and environmental noise, it also increases the complexity of heart rate data collection and analysis. Therefore, it is necessary to provide a portable heart rate anomaly warning system based on edge computing to optimize the above - mentioned technical problems. Summary of the Invention

[0003] The present invention relates to the technical field of heart rate monitoring, and particularly relates to a portable heart rate anomaly warning system based on edge computing. To solve the above - mentioned technical problems, the present invention provides a portable heart rate anomaly warning system based on edge computing, which combines a micro - sensor array, an edge computing node, and a multi - modal data fusion algorithm to achieve real - time collection, analysis, and anomaly warning functions for heart rate data.

[0004] A portable heart rate anomaly warning method provided by the present invention includes the following steps: collecting the physiological signals of a user through a micro - sensor array and transmitting the collected signals to an edge computing node; in the edge computing node, using a signal pre - processing unit to perform filtering and denoising operations on the original signals to extract effective heart rate features; based on the extracted heart rate features, using a multi - modal data fusion algorithm to integrate the output data of multiple sensors to generate comprehensive heart rate data; inputting the comprehensive heart rate data into an anomaly detection model, which is a machine - learning classifier trained based on historical heart rate data, to determine whether the current heart rate is in an abnormal state; if an abnormal heart rate is detected, a warning signal is sent through a local alarm module, and the abnormal data is uploaded to a cloud server for further analysis and storage; if no abnormality is detected, continue to collect physiological signals in the next cycle and repeat the above process.

[0005] The present invention also provides a portable heart rate anomaly warning system based on edge computing, which is applied to the above - mentioned method. The system includes the following components: a signal collection module, a signal pre - processing module, a data fusion module, an anomaly detection module, a local alarm module, and a cloud interaction module.

[0006] The signal acquisition module consists of multiple micro sensors, which are connected to the signal acquisition unit through a flexible circuit board and are used to collect the user's physiological signals in real time. Each sensor includes a photoplethysmography sensor and an acceleration sensor, which are used to collect optical signals related to heart rate and motion interference signals respectively. The signal acquisition module establishes a wireless communication connection with the edge computing node through the low-power Bluetooth protocol to ensure the stability and real-time nature of data transmission.

[0007] The signal preprocessing module is deployed inside the edge computing node and includes a digital signal processor and a set of preset filtering algorithms. The digital signal processor processes the received raw signals in segments. First, it uses a band-pass filter to remove high-frequency noise and low-frequency drift, and then extracts the main frequency components of the heart rate signal through wavelet transform. In addition, the signal preprocessing module also eliminates motion interference signals through an adaptive filtering algorithm, thereby improving the accuracy of heart rate features.

[0008] The data fusion module receives multi-channel output data from the signal preprocessing module and integrates these data using a multi-modal data fusion algorithm. The multi-modal data fusion algorithm is based on the Kalman filtering principle and generates comprehensive heart rate data by dynamically adjusting the weight coefficients of each sensor's data. This algorithm can automatically optimize the weight allocation according to environmental changes and the user's motion state, thereby enhancing the reliability of the comprehensive heart rate data.

[0009] The anomaly detection module includes a classifier based on support vector machines. This classifier establishes the classification boundary between normal heart rate and abnormal heart rate through learning historical heart rate data. During operation, the anomaly detection module inputs the comprehensive heart rate data into the classifier and determines whether the current heart rate is abnormal by calculating the classification score. If the classification score exceeds the preset threshold, it is determined as an abnormal heart rate; otherwise, it is determined as a normal heart rate.

[0010] The local alarm module is deployed on the portable device and includes a vibration motor and an LED indicator. When the anomaly detection module determines that the heart rate is abnormal, the local alarm module generates continuous vibration by controlling the vibration motor and lights up the LED indicator to remind the user. At the same time, the local alarm module generates voice prompt information through a voice synthesis chip to inform the user of the current heart rate status and the recommended measures to be taken.

[0011] The cloud interaction module is responsible for uploading abnormal data to the cloud server and downloading updated model parameters from the cloud server. The cloud interaction module establishes a communication connection with the cloud server through Wi-Fi or a cellular network and uses the HTTPS protocol to ensure the security of data transmission. The cloud server conducts in-depth analysis on the received abnormal data and feeds back the analysis results to the portable device for the user to view the detailed report.

[0012] Compared with related technologies, a portable heart rate anomaly warning system based on edge computing provided by the present invention has the following characteristics: The present invention collects the physiological signals of users through a micro sensor array, and uses a signal preprocessing module to filter and denoise the original signals, so as to extract effective heart rate features; Then, a data fusion module integrates multi-channel sensor data to generate comprehensive heart rate data; Subsequently, an anomaly detection module classifies and judges the comprehensive heart rate data to detect abnormal heart rates in a timely manner; Finally, a local alarm module issues a warning signal and uploads the abnormal data to a cloud server for further analysis. The present invention realizes real-time data processing through edge computing nodes, reduces data transmission delay, and improves monitoring efficiency; At the same time, through multi-modal data fusion algorithms and anomaly detection models, the accuracy and reliability of heart rate monitoring are improved, providing users with more accurate health warning services. Description of the Drawings

[0013] Figure 1 It is the overall structural block diagram of the system of the present invention, showing the connection relationships among a signal acquisition module, a signal preprocessing module, a data fusion module, an anomaly detection module, a local alarm module, and a cloud interaction module.

[0014] Figure 2 It is the composition schematic diagram of the signal acquisition module, including a micro sensor array and its wireless communication connection method with an edge computing node.

[0015] Figure 3 It is the working flow chart of the signal preprocessing module, detailing the processing process of the original signal after band-pass filtering, wavelet transform, and adaptive filtering.

[0016] Figure 4 It is the algorithm flow chart of the data fusion module, showing the process of the multi-modal data fusion algorithm adjusting the weights of multi-channel sensor data based on the Kalman filter principle.

[0017] Figure 5 It is the working principle diagram of the classifier of the anomaly detection module, illustrating how a support vector machine classifier establishes a classification boundary and judges the current heart rate state through historical heart rate data.

[0018] Figure 6 It is the function implementation diagram of the local alarm module, showing the collaborative working mode of a vibration motor, an LED indicator, and a voice synthesis chip in anomaly warning.

[0019] Figure 7 It is the data transmission schematic diagram of the cloud interaction module, depicting the communication connection and data upload / download process between a portable device and a cloud server.

[0020] Figure 8It is the overall timing diagram of the system operation, showing the complete working cycle from signal acquisition to abnormal warning and data upload.

[0021] The attached drawing reference numerals are as follows:

[0022] 1. Signal acquisition module; 2. Signal preprocessing module; 3. Data fusion module; 4. Abnormal detection module; 5. Local alarm module; 6. Cloud interaction module; 7. Micro sensor array; 8. Edge computing node. Specific implementation manners

[0023] The present invention provides a portable heart rate abnormal warning system based on edge computing, and its overall structure is as Figure 1 shown. The system includes a signal acquisition module 1, a signal preprocessing module 2, a data fusion module 3, an abnormal detection module 4, a local alarm module 5 and a cloud interaction module 6. These modules cooperate through hardware circuits and software algorithms to realize the functions of real-time acquisition, processing, analysis and warning of heart rate data. The specific implementation manners of the system will be described in detail below with reference to the attached drawings.

[0024] The signal acquisition module 1 is composed of a micro sensor array 7, as Figure 2 shown. The micro sensor array 7 includes multiple sensor units, and each unit integrates a photoplethysmography sensor and an acceleration sensor. The photoplethysmography sensor is used to collect optical signals related to the heart rate, while the acceleration sensor is used to monitor the user's motion state to exclude motion interference. The micro sensor array 7 is connected to the signal acquisition unit through a flexible circuit board, and the signal acquisition unit establishes a wireless communication connection with the edge computing node 8 through a low-power Bluetooth protocol. This connection method ensures the stability and real-time performance of data transmission, while reducing power consumption, and is suitable for long-term wearing and use.

[0025] The signal preprocessing module 2 is deployed inside the edge computing node 8, and its working process is as Figure 3 shown. The signal preprocessing module 2 includes a digital signal processor and a set of preset filtering algorithms. When the original signal is transmitted from the signal acquisition module 1 to the edge computing node 8, the digital signal processor first performs segmented processing on the received signal. The segmented signals are successively passed through a band-pass filter to remove high-frequency noise and low-frequency drift, and then the main frequency components of the heart rate signal are extracted through wavelet transform. In addition, the signal preprocessing module 2 also adopts an adaptive filtering algorithm, using the motion signal collected by the acceleration sensor as a reference input to dynamically eliminate the motion interference signal, thereby improving the accuracy of the heart rate characteristics. This process ensures the quality of the basic data for subsequent data analysis.

[0026] The data fusion module 3 receives multiplexed output data from the signal preprocessing module 2 and integrates these data using a multimodal data fusion algorithm. The algorithm flow is as Figure 4 shown. The multimodal data fusion algorithm is based on the Kalman filtering principle and generates comprehensive heart rate data by dynamically adjusting the weight coefficients of multiplexed sensor data. Specifically, the Kalman filter automatically optimizes the weight allocation of each sensor data according to environmental changes and the user's motion state. For example, when the user is in a stationary state, the data weight of the photoplethysmography sensor is higher; while when the user is moving, the data weight of the acceleration sensor will increase accordingly. This dynamic adjustment mechanism significantly improves the reliability of the comprehensive heart rate data and provides high-quality input data for subsequent anomaly detection.

[0027] The anomaly detection module 4 includes a classifier based on a support vector machine. Its working principle is as Figure 5 shown. The support vector machine classifier establishes the classification boundary between normal heart rate and abnormal heart rate by learning historical heart rate data. During operation, the anomaly detection module 4 inputs the comprehensive heart rate data into the classifier and determines whether the current heart rate is abnormal by calculating the classification score. If the classification score exceeds the preset threshold, it is determined as an abnormal heart rate; otherwise, it is determined as a normal heart rate. This classification process depends on the training results of historical data, so the performance of the system gradually improves as the usage time increases.

[0028] The local alarm module 5 is deployed on the portable device. Its function implementation is as Figure 6 shown. The local alarm module 5 includes a vibration motor, an LED indicator, and a voice synthesis chip. When the anomaly detection module 4 determines that the heart rate is abnormal, the local alarm module 5 generates continuous vibration by controlling the vibration motor and lights up the LED indicator to alert the user. At the same time, the voice synthesis chip generates voice prompt information to inform the user of the current heart rate status and the recommended measures to take. This multi-mode alarm method ensures that users can receive warning information in a timely manner in different scenarios.

[0029] The cloud interaction module 6 is responsible for uploading abnormal data to the cloud server and downloading updated model parameters from the cloud server. Its data transmission process is as Figure 7 shown. The cloud interaction module 6 establishes a communication connection with the cloud server through Wi-Fi or a cellular network and uses the HTTPS protocol to ensure the security of data transmission. The cloud server conducts in-depth analysis on the received abnormal data and feeds back the analysis results to the portable device for the user to view detailed reports. In addition, the cloud server can also update the model parameters of the support vector machine classifier according to the latest data, thereby improving the detection accuracy of the system.

[0030] The overall operation timing of the system is as Figure 8As shown in the figure. First, the signal acquisition module 1 collects the user's physiological signals in real time through the micro-sensor array 7 and transmits the collected signals to the edge computing node 8. In the edge computing node 8, the signal preprocessing module 2 performs filtering and denoising operations on the original signals to extract effective heart rate features. Subsequently, the data fusion module 3 uses a multi-modal data fusion algorithm to integrate multi-channel sensor data to generate comprehensive heart rate data. The comprehensive heart rate data is input into the anomaly detection module 4, and a support vector machine classifier is used to determine whether the current heart rate is abnormal. If an abnormal heart rate is detected, the local alarm module 5 issues a warning signal and uploads the abnormal data to the cloud server for further analysis and storage. If no abnormality is detected, the physiological signals of the next cycle are continuously collected and the above process is repeated.

[0031] In practical applications, this system can be widely used in fields such as personal health management, sports monitoring, and medical monitoring. For example, when the user wears this device in daily life, the system can monitor their heart rate status in real time and issue a warning in a timely manner when an abnormality occurs. For athletes, this system can help them monitor heart rate changes during training and avoid health risks caused by overtraining. In addition, in the medical monitoring scenario, medical staff can view the abnormal data and analysis reports of patients through the cloud server, so as to formulate a more accurate treatment plan.

[0032] In order to better enable relevant personnel in the technical field to fully understand and implement the present invention, the specific implementation principle of the present invention is further supplemented and explained below in combination with a specific application scenario.

[0033] In the application scenario of personal health management, when the user wears the portable heart rate anomaly warning system, first, the micro-sensor array 7 in the signal acquisition module 1 collects the user's physiological signals in real time. As Figure 2 shown, the micro-sensor array 7 includes multiple sensor units, and each unit integrates a photoplethysmography sensor and an acceleration sensor. The photoplethysmography sensor emits a specific wavelength of light beam to penetrate the skin surface layer and receives the reflected light signal, thereby obtaining the optical signal related to the heart rate; while the acceleration sensor is used to monitor the user's motion state to exclude the influence of motion interference on the heart rate data. The micro-sensor array 7 transmits the collected original signals to the edge computing node 8 through the low-power Bluetooth protocol. This wireless communication method effectively reduces power consumption and ensures the stability and real-time nature of data transmission, making it possible to wear for a long time.

[0034] Subsequently, the signal preprocessing module 2 processes the received original signals inside the edge computing node 8. As Figure 3As shown, the digital signal processor first segments the original signal to facilitate subsequent filtering. The segmented signal then passes through a band-pass filter in sequence to remove high-frequency noise and low-frequency drift, which may stem from ambient light interference or drift errors in the sensor itself. Next, the signal undergoes wavelet transform to extract the dominant frequency components, a step that can isolate the key frequency information related to heart rate characteristics. Additionally, the signal preprocessing module 2 also employs an adaptive filtering algorithm, using the motion signal collected by the acceleration sensor as the reference input to dynamically eliminate the interference signals generated by user movement. This comprehensive filtering method significantly improves the accuracy of heart rate characteristics and provides high-quality basic data for subsequent data fusion.

[0035] Next, the data fusion module 3 receives the multiplexed output data from the signal preprocessing module 2 and integrates it based on the Kalman filtering principle. As Figure 4 shown, the multi-modal data fusion algorithm dynamically adjusts the weight coefficients of the multiplexed sensor data to generate the integrated heart rate data. For example, in the user's stationary state, since the data from the photoplethysmography sensor is less affected by movement interference, its weight coefficient is higher; while when the user is moving, the weight of the acceleration sensor data will increase accordingly to make up for the deficiencies of the photoplethysmography sensor in a high-dynamic environment. The Kalman filter automatically optimizes the weight allocation of each sensor data according to environmental changes and the user's motion state, thereby enhancing the reliability of the integrated heart rate data. This mechanism ensures that regardless of the user's state, the system can generate accurate heart rate data for subsequent analysis.

[0036] In the anomaly detection stage, the anomaly detection module 4 inputs the integrated heart rate data into a classifier based on support vector machines, as Figure 5 shown. The support vector machine classifier establishes the classification boundary between normal and abnormal heart rates through learning the historical heart rate data. During operation, the classifier calculates the classification score of the current heart rate data and compares it with a preset threshold. If the classification score exceeds the threshold, it is determined as an abnormal heart rate; otherwise, it is determined as a normal heart rate. Since the performance of the classifier depends on the training results of historical data, as the system is used over time, the classifier can gradually optimize its judgment ability, thereby improving the accuracy of anomaly detection.

[0037] When the anomaly detection module 4 determines that the heart rate is abnormal, the local alarm module 5 immediately activates the early warning function. As Figure 6As shown, the local alarm module 5 generates continuous vibrations through a vibration motor and lights up the LED indicator to alert the user. At the same time, the speech synthesis chip generates voice prompt information to inform the user of the current heart rate status and the recommended measures. For example, when a tachycardia is detected, the voice prompt may include guiding content such as "Your heart rate is too high. Please rest immediately." This multi-mode alarm method ensures that users can receive warning messages in a timely manner in different scenarios, effectively attracting the user's attention whether in a noisy environment or a quiet environment.

[0038] Meanwhile, the cloud interaction module 6 is responsible for uploading abnormal data to the cloud server and downloading updated model parameters from the cloud server. As Figure 7 shown, the cloud interaction module 6 establishes a communication connection with the cloud server through Wi-Fi or a cellular network and uses the HTTPS protocol to ensure the security of data transmission. The cloud server conducts in-depth analysis on the received abnormal data, generates a detailed health report, and feeds back the analysis results to the portable device for the user to view. In addition, the cloud server can also update the model parameters of the support vector machine classifier based on the latest data, thereby further improving the detection accuracy of the system. Through cloud interaction, the system realizes the long-term storage and analysis capabilities of data, providing users with more comprehensive health management services.

[0039] The overall operation timing of the system is as Figure 8 shown. The signal acquisition module 1 continuously acquires the user's physiological signals and transmits the acquired signals to the edge computing node 8. In the edge computing node 8, the signal preprocessing module 2 performs filtering and denoising operations on the original signals to extract effective heart rate features; the data fusion module 3 uses a multi-modal data fusion algorithm to integrate multi-channel sensor data to generate comprehensive heart rate data; the anomaly detection module 4 determines whether the current heart rate is abnormal through a support vector machine classifier. If an abnormal heart rate is detected, the local alarm module 5 issues a warning signal and uploads the abnormal data to the cloud server for further analysis and storage; if no abnormality is detected, the system continues to acquire the physiological signals of the next cycle and repeats the above process.

[0040] Through the above steps, this system can achieve efficient and accurate heart rate monitoring and abnormal warning functions in personal health management. For example, when the user wears this device in daily life, the system can monitor their heart rate status in real time and issue a warning in a timely manner when an abnormality occurs, helping the user take necessary countermeasures. For athletes, this system can monitor heart rate changes during training to avoid health risks caused by overtraining. In addition, in the medical monitoring scenario, medical staff can view the abnormal data and analysis reports of patients through the cloud server, so as to formulate a more accurate treatment plan. The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A portable abnormal heart rate warning system based on edge computing, characterized in that: It comprises a signal acquisition module (1), a signal preprocessing module (2), a data fusion module (3), an anomaly detection module (4), a local alarm module (5) and a cloud interaction module (6), wherein: The signal acquisition module (1) is used to collect physiological signals of the user through a micro sensor array (7) and transmit the collected signals to an edge computing node (8); The signal preprocessing module (2) is deployed inside the edge computing node (8) and is used to filter and denoise the received original signal to extract effective heart rate features; The data fusion module (3) is used to receive the multi-channel output data from the signal preprocessing module (2), and integrate the multi-channel sensor data using a multi-modal data fusion algorithm to generate comprehensive heart rate data; The anomaly detection module (4) includes a classifier based on a support vector machine, which is used to input the comprehensive heart rate data into the classifier to determine whether the current heart rate is in an abnormal state; The local alarm module (5) is used to send out a warning signal when an abnormal heart rate is detected; The cloud interaction module (6) is used to upload abnormal data to the cloud server and download updated model parameters from the cloud server.

2. The portable abnormal heart rate warning system according to claim 1, characterized in that: The signal acquisition module (1) is composed of a plurality of micro sensors, each of which includes a photoplethysmography sensor and an acceleration sensor. The photoplethysmography sensor is used to collect optical signals related to the heart rate, and the acceleration sensor is used to collect motion interference signals.

3. The portable abnormal heart rate warning system according to claim 1, characterized in that: The signal preprocessing module (2) includes a digital signal processor and a set of preset filtering algorithms. The digital signal processor sequentially performs bandpass filtering, wavelet transform and adaptive filtering operations on the received original signal to remove high-frequency noise, low-frequency drift and motion interference signals.

4. The portable abnormal heart rate warning system according to claim 1, characterized in that: The data fusion module (3) uses a multimodal data fusion algorithm based on the Kalman filter principle to dynamically adjust the weight coefficients of multi-channel sensor data to generate comprehensive heart rate data.

5. The portable abnormal heart rate warning system according to claim 1, characterized in that: The support vector machine classifier in the anomaly detection module (4) establishes the classification boundary between normal heart rate and abnormal heart rate by learning the historical heart rate data, and determines whether the current heart rate is abnormal by calculating the classification score.

6. The portable abnormal heart rate warning system according to claim 1, characterized in that: The local alarm module (5) comprises a vibration motor, an LED indicator light and a speech synthesis chip. When an abnormal heart rate is detected, the vibration motor generates continuous vibration, the LED indicator lightens, and the speech synthesis chip generates a speech prompt message.

7. The portable abnormal heart rate warning system according to claim 1, characterized in that: The cloud interaction module (6) establishes a communication connection with the cloud server via Wi-Fi or cellular network, and uses the HTTPS protocol to ensure the security of data transmission.

8. A portable heart rate abnormality warning method based on edge computing, characterized in that: The following steps are involved: S1: collecting physiological signals of the user through a micro sensor array (7) and transmitting the collected signals to an edge computing node (8); S2: In the edge computing node (8), the signal preprocessing module (2) is used to filter and denoise the original signal to extract effective heart rate features; S3: using the data fusion module (3) to integrate the multi-channel sensor data to generate comprehensive heart rate data; S4: inputting the comprehensive heart rate data into the abnormality detection module (4) and determining whether the current heart rate is abnormal through a support vector machine classifier; S5: If an abnormal heart rate is detected, a warning signal is issued through the local alarm module (5), and the abnormal data is uploaded to the cloud server; if no abnormality is detected, the physiological signal of the next cycle continues to be collected and the above process is repeated.

9. The portable abnormal heart rate warning method according to claim 8, characterized in that: In step S3, the data fusion module (3) uses a multimodal data fusion algorithm based on the Kalman filter principle to dynamically adjust the weight coefficient of each sensor data according to environmental changes and user motion status to generate comprehensive heart rate data.

Images