Atrial fibrillation detection and warning method and electrocardiogram data relay device

Abstract

The present application relates to the medical health technical field, propose a kind of atrial fibrillation detection early warning method and electrocardiogram data relay device, by training obtain a light atrial fibrillation detection model, in convolution layer alternately use conventional convolution and depth separable convolution, whether the real-time detection of collected electrocardiogram waveform is abnormal in offline state can be realized, to issue corresponding alarm prompt.The main body of the electrocardiogram data relay device is composed of main controller module, bluetooth module, OTG module, storage module, cellular network and WIFI module, power supply module, key module, LCD module, LED module, while embedding the above light atrial fibrillation detection model into main controller, and the electrocardiogram data received by bluetooth and OTG module is collected, electrocardiogram data is sent to remote server for detection by WIFI or cellular network, and the electrocardiogram data can also be placed in storage module, to reduce the workload of electrocardiogram acquisition equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical health, in particular to a method for detecting and warning atrial fibrillation and an electrocardiogram data relay device. BACKGROUND

[0002] With the development of intelligent technology, wearable devices are inevitable. Wearable technology has developed into a wide-ranging industry, including not only fitness and health devices, but also smartwatches, motion tracking devices, and other embeddable devices. At the same time, due to the aging of the population and the shortage of medical resources, medical wearable monitoring devices have also attracted more and more attention from the public. The medical wearable monitoring devices on the market mainly include smart bracelets and smart watches, which have the characteristics of strong operability, portability, and attractive appearance. They mainly monitor physiological parameters such as electrocardiogram, blood oxygen, respiration, and pulse in real time. However, due to the particularity of medical wearable monitoring devices, they require real-time monitoring (large data volume), high accuracy of data, and other characteristics, so continuous, efficient, and stable data transmission is particularly important, and monitoring and early warning are also required.

[0003] In the prior art, data from an electrocardiogram collection device is mostly transmitted using Bluetooth. However, due to the popularity of wireless devices in daily household use, the anti-interference performance of Bluetooth data transmission is required to be high. Therefore, a single data transmission method using Bluetooth cannot meet the high stability requirements of household medical devices.

[0004] In the prior art, when data from an electrocardiogram collection device is sent to a server, one of cellular network (4G / 5G) or WIFI mode, i.e. single mode, is mostly used. This is obviously not comprehensive enough.

[0005] In the prior art, to detect whether the electrocardiogram waveform is normal, the electrocardiogram data needs to be sent to the server for detection and then fed back to the device. This requires online network status, i.e. it is impossible to detect whether the electrocardiogram waveform is normal in offline state.

[0006] In the Chinese invention patent with publication date of May 6, 2022 and publication number CN114449496A, an electrocardiogram collection device uses a single mode transmission method to transmit the collected electrocardiogram signal to a relay device. The relay device can only send the data to a remote server for electrocardiogram abnormality detection after receiving the data sent by the electrocardiogram collection device. If the electrocardiogram detection is abnormal, the remote server needs to send data to the device for corresponding alarm prompt. Therefore, it cannot realize real-time detection of the electrocardiogram signal. SUMMARY

[0007] To this end, the technical problem to be solved by the present application is to provide an atrial fibrillation detection and early warning method and an electrocardio data relay device to overcome the problems of unstable Bluetooth single-mode transmission, single data transmission mode and inability to detect whether the electrocardio signal is abnormal in an offline state in the prior art.

[0008] To solve the above technical problems, the present application provides an atrial fibrillation detection and early warning method, comprising:

[0009] selecting a group of electrocardio samples in an electrocardio signal database and inputting the electrocardio samples into a detection network to obtain R peak positions of the electrocardio samples;

[0010] calculating time distances RR intervals between adjacent R peaks according to the R peak positions, and a previous RR interval and a subsequent RR interval of any one R peak forming a two-channel RR interval wave, if a fluctuation range of the two-channel RR interval wave is greater than or equal to a threshold value, then marking "1" to represent that atrial fibrillation occurs, and if the fluctuation range of the two-channel RR interval wave is less than the threshold value, then marking "0" to represent that atrial fibrillation does not occur, and obtaining a true label of the two-channel RR interval wave;

[0011] inputting the two-channel RR interval wave into a lightweight atrial fibrillation detection model, alternately using a regular convolution and a depth separable convolution unit to extract features of the two-channel RR interval wave, and obtaining a feature map of the two-channel RR interval wave; wherein the regular convolution and the depth separable convolution unit use a lightweight attention module in the second layer to improve the feature extraction capability of the lightweight atrial fibrillation detection model;

[0012] processing the feature map of the two-channel RR interval wave using a sigmoid activation function to obtain a detection result;

[0013] fitting and training the detection result and the true label of the two-channel RR interval wave to obtain a lightweight atrial fibrillation detection model with the smallest loss function value.

[0014] Preferably, the lightweight attention module is implemented using a CBAM attention mechanism.

[0015] Preferably, the lightweight atrial fibrillation detection model presents a left-right symmetric structure, and on one side of each level, there are four convolution units, including a one-dimensional convolution, a batch normalization layer and a leaky rectified linear unit.

[0016] Preferably, the alternately used regular convolution and depth separable convolution unit extracts features of the two-channel RR interval wave, and the calculation formula of the depth separable convolution unit is:

[0017]

[0018] wherein D KThe size of the convolution kernel, C in and C out are the number of channels of input and output, respectively.

[0019] The application also provides an electrocardio data relay device, comprising:

[0020] a housing, a main controller module, a Bluetooth module, an OTG module, a storage module, a cellular network and WIFI module, a power supply module arranged inside the housing, and a key module, an LCD module, and an LED module arranged on the outer surface of the housing in sequence.

[0021] The main controller module stores a computer program, which, when executed by the main control module, implements the steps of the atrial fibrillation detection and early warning method described above, including a lightweight detection unit for detecting atrial fibrillation of the electrocardio signal in an offline state.

[0022] The Bluetooth module and the OTG module are used to receive the electrocardio signal sent by the electrocardio acquisition device.

[0023] The storage module is used to store the electrocardio signal.

[0024] The cellular network and WIFI module are used to send the electrocardio signal to a remote server for atrial fibrillation detection in an online network state.

[0025] The key module is used to send a control signal to the main control module.

[0026] The LCD module is used to display the electrocardio waveform after atrial fibrillation detection.

[0027] The LED module is used for abnormal alarm of electrocardio signal atrial fibrillation detection.

[0028] The power supply module is used to power the electrocardio data relay device.

[0029] Preferably, the Bluetooth module includes an NFC unit for obtaining the Bluetooth mac address of the electrocardio acquisition device through NFC identification, pairing the Bluetooth of the electrocardio data relay device and the electrocardio acquisition device, and transmitting the electrocardio signal.

[0030] Preferably, the key module includes a first key and a second key, the first key is used to control the on-off of the electrocardio data relay device, and the second key is used to turn on and off the LCD display screen.

[0031] Preferably, the power supply module uses a lithium battery for power supply and an additional charging circuit for charging the battery.

[0032] Preferably, the LCD module can also display the battery level information of the electrocardio data relay device.

[0033] The above technical solutions of the present application have the following advantages compared with the prior art:

[0034] The atrial fibrillation detection and early warning method provided by the present application trains a lightweight atrial fibrillation detection model, uses regular convolution and depth separable convolution alternately in the convolution layer, and compared with regular convolution, the one-dimensional depth separable convolution reduces the parameter quantity and speeds up the model inference speed.

[0035] In addition, the electrocardio data relay device provided by the present application receives electrocardio data of an electrocardio data collection device by using two ways of a Bluetooth module and an OTG module, so that the electrocardio data can be continuously and stably relayed; a storage module is arranged on the electrocardio data relay device, and the electrocardio data can be saved in the relay device, thereby reducing the workload of the electrocardio data collection device; when the electrocardio data is sent to a remote server, the networking mode of the electrocardio data relay device provided by the present application has two modes of a cellular version and a WIFI version, and a trained lightweight atrial fibrillation detection model is embedded in the electrocardio data relay device, so that whether the collected electrocardio waveform is abnormal can be detected in real time in an offline state. BRIEF DESCRIPTION OF DRAWINGS

[0036] In order to make the content of the present application more easily understood, the present application will be further described in detail below according to specific embodiments of the present application and in combination with the drawings, in which

[0037] Figure 1 The atrial fibrillation detection and early warning method provided by the present application;

[0038] Figure 2 The network structure diagram of the lightweight atrial fibrillation detection model;

[0039] Figure 3 The system block diagram of the electrocardio data relay device provided by the present application. DETAILED DESCRIPTION

[0040] The present application will be further described below in combination with the drawings and specific embodiments, so that those skilled in the art can better understand the present application and implement it, but the embodiments are not limiting to the present application.

[0041] The core of the present application is to provide an atrial fibrillation detection and early warning method and an electrocardio data relay device, which realize multi-mode transmission of electrocardio data, continuous and stable relay data, and real-time detection of whether the electrocardio waveform is abnormal.

[0042] Referring to Figure 1 as shown, Figure 1The atrial fibrillation detection and early warning method provided by the embodiment of the application comprises:

[0043] Step S101: selecting a group of electrocardio samples in an electrocardio signal database and inputting the electrocardio samples into a detection network to obtain R peak positions of the electrocardio samples;

[0044] The algorithm used by the detection network can be any one of a PT algorithm and an end-to-end R peak detection algorithm;

[0045] Step S102: calculating time distances RR intervals between adjacent R peaks according to the R peak positions;

[0046] The previous RR interval and the next RR interval of any one R peak form a two-channel RR interval wave, if the fluctuation range of the two-channel RR interval wave is greater than or equal to a threshold, then it is marked as "1" to indicate that atrial fibrillation occurs, and if the fluctuation range of the two-channel RR interval wave is less than the threshold, then it is marked as "0" to indicate that atrial fibrillation does not occur, and a true label of the two-channel RR interval wave is obtained;

[0047] Step S103: detecting a segment where atrial fibrillation occurs in the RR interval wave by using a network, inputting the two-channel RR interval wave into a lightweight atrial fibrillation detection model, alternately using a regular convolution and a depth separable convolution unit to extract features of the two-channel RR interval wave, and obtaining a feature map of the two-channel RR interval wave;

[0048] The convolution unit uses a lightweight attention module in the second layer to improve the feature extraction capability of the lightweight atrial fibrillation detection model;

[0049] Referring to Figure 2 , Figure 2 The network structure diagram of the lightweight atrial fibrillation detection model, the network presents a left-right symmetrical structure, on one side of each level of the network, there are four convolution units, the convolution unit is composed of a one-dimensional convolution, a batch normalization (BN) layer and a leaky rectified linear unit (LRELU), as shown in the right rectangular box of Figure 2 ; the filter size of the one-dimensional convolution in all convolution units is 24, and the kernel size of the one-dimensional convolution in all convolution units at levels 0 to 2 is 3; each downsampling operation reduces the length of its input feature map by half, and correspondingly, each upsampling operation doubles the length of its input feature map;

[0050] The lightweight attention module is implemented by using a CBAM attention mechanism;

[0051] Compared with a regular convolution, the parameter amount of the one-dimensional depth separable convolution is reduced, and the formula of the regular convolution unit is:

[0052]

[0053] The calculation formula of the one-dimensional depth separable convolution unit is:

[0054]

[0055] Wherein, D K is the size of the convolution kernel, C in and C out are the number of input and output channels respectively, and W out is the length of the output feature map.

[0056] Step S104: using a sigmoid activation function to process the feature map of the two-channel RR interval wave to obtain a detection result.

[0057] Step S105: fitting training the detection result with the true label of the two-channel RR interval wave to obtain a lightweight atrial fibrillation detection model with the minimum loss function value.

[0058] The trained lightweight atrial fibrillation detection model performs segmented atrial fibrillation detection on the electrocardio signal in an overlapping sliding window manner, and the specific steps are as follows:

[0059] Step S201: inputting the ECG signal into the detection model, setting the sliding window width to 6.48s, and setting the overlapping time length to 2.28s.

[0060] Step S202: down-sampling the ECG signal to 100HZ to obtain the R peak position.

[0061] Step S203: obtaining the previous RR interval and the next RR interval of each R peak, and generating a two-channel RR interval wave.

[0062] Step S204: inputting the two-channel RR interval wave into the lightweight atrial fibrillation detection model, if the atrial fibrillation phenomenon exists in the detection waveform, outputting “1”, otherwise outputting “0”.

[0063] Step S205: when the ECG signal is not ended, repeating the above steps.

[0064] The atrial fibrillation detection and early warning method provided in the embodiment of the application trains a lightweight atrial fibrillation detection model, alternately uses a conventional convolution and a depth separable convolution in the convolution layer, and compared with the conventional convolution, the one-dimensional depth separable convolution reduces the parameter quantity and speeds up the model inference speed.

[0065] According to the above embodiment, the embodiment provides an electrocardio data relay device;

[0066] Referring to Figure 3 , Figure 3The system block diagram of the electrocardio data relay device provided by the application, the device body is composed of a main controller module, a Bluetooth module, an OTG module, a storage module, a cellular network and WIFI module, a power supply module, a key module, an LCD module and an LED module;

[0067] The main controller module inside the device is connected with the Bluetooth module, the OTG module, the storage module, the cellular network and WIFI module, the power supply module, the key module, the LCD module and the LED module, and is used for controlling the interaction between the modules.

[0068] The main controller module comprises a lightweight detection unit, the lightweight detection unit stores a computer program, and the computer program is executed by the main controller module to realize the steps of any one of the atrial fibrillation detection and early warning methods described in the above embodiments, so as to facilitate the atrial fibrillation detection on the electrocardio signal in an offline state.

[0069] The electrocardio data relay device further comprises:

[0070] The shell, the main controller module, the Bluetooth module, the OTG module, the storage module, the cellular network and WIFI module and the power supply module arranged inside the shell, and the key module, the LCD module and the LED module arranged on the outer surface of the shell in sequence.

[0071] The SIM card slot arranged on the shell of the electrocardio data relay device is used for inserting a SIM card and transmitting the electrocardio signal by using a cellular network.

[0072] The Bluetooth module and the OTG module are used for receiving the electrocardio signal sent by the electrocardio acquisition device.

[0073] The Bluetooth module comprises an NFC unit, which is used for obtaining the Bluetooth mac address of the electrocardio acquisition device through NFC identification, pairing the Bluetooth of the electrocardio data relay device with the electrocardio acquisition device, and transmitting the electrocardio signal.

[0074] The storage module is used for storing the electrocardio signal.

[0075] The cellular network and WIFI module are used for sending the electrocardio signal to a remote server for atrial fibrillation detection in a network online state.

[0076] The key module is used for sending a control signal to the main control module.

[0077] The key module comprises a first key and a second key, the first key is used for controlling the on-off of the electrocardio data relay device, and the second key is used for opening and closing the LCD display screen.

[0078] An LCD module is arranged to display the ECG waveform after the atrial fibrillation detection and the battery power information of the ECG data relay device.

[0079] An LED module is arranged to alarm the ECG signal atrial fibrillation detection abnormality, and the LED light flashes to alarm when the ECG waveform is detected to have the atrial fibrillation, indicating the ECG waveform abnormality.

[0080] A power supply module is arranged to supply power to the ECG data relay device.

[0081] The power supply module is powered by a lithium battery and has a charging circuit to charge the battery.

[0082] The specific steps of the real-time monitoring of the ECG waveform by the atrial fibrillation detection and early warning method and the ECG data relay device are as follows:

[0083] In step S401, when starting to use, the ECG collection equipment is close to the Bluetooth module of the ECG signal relay device, the NFC unit in the Bluetooth module reads the Bluetooth mac address of the ECG collection equipment at a close distance, the ECG collection equipment and the ECG signal relay device are quickly Bluetooth paired successfully, the ECG data transmission is performed, or the OTG mode can also be used to insert the USB male port of the ECG collection equipment into the USB female port of the ECG data relay device to perform the ECG data transmission.

[0084] In step S402, the ECG data sent by the ECG collection equipment is stored in the storage module of the ECG data relay device.

[0085] In step S403, the second key is opened to turn on the LCD display screen, the battery power information is displayed on the screen, the display screen can be turned off by pressing the key to save the power when the display screen is not needed, and the main controller module of the ECG data relay device starts to detect the ECG waveform, which is divided into two cases of network online and offline, and the detection process is as follows:

[0086] In the network online state, the ECG data relay device can select one of WIFI or cellular network (4G / 5G) to connect the network to transmit the ECG data to the remote server, when the remote server detects the ECG data abnormality, the information is fed back to the main controller module, the main controller module controls the LCD display screen to display the ECG waveform after the atrial fibrillation detection, if the atrial fibrillation phenomenon is detected, the LED light flashes red light to alarm the ECG waveform detection abnormality, and if the atrial fibrillation phenomenon is not detected, the LED light does not flash.

[0087] If the cellular network is selected to transmit the ECG data, the SIM card needs to be inserted into the SIM card slot of the ECG data relay device.

[0088] In the offline state, the ECG data relay device main controller module controls the lightweight detection unit to detect atrial fibrillation of the received ECG signal, and sends the detection result to the display module, and the LCD display screen displays the ECG waveform after atrial fibrillation detection, if the atrial fibrillation phenomenon is detected, the LED lamp flashes red light, to prompt the ECG waveform detection abnormal alarm, if the atrial fibrillation phenomenon is not detected, the LED lamp does not flash.

[0089] The atrial fibrillation detection and early warning method and the ECG data relay device provided by the application embed a lightweight atrial fibrillation detection model in the ECG data relay device, which guarantees the real-time monitoring function of the ECG data relay device, so that the ECG data relay device can run continuously and stably.

[0090] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system, or a computer program product. Therefore, the application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the application can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0091] The application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0092] These computer program instructions can also be stored in a computer readable storage medium that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction devices, which implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0093] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide processes for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0094] Obviously, the above embodiments are only examples for clearly illustrating the present application, and are not intended to limit the embodiments. Based on the above description, other different forms of changes or variations can be made by those skilled in the art. Here, all the embodiments are not required to be exhausted, and the obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. A method for detecting and warning of atrial fibrillation, characterized in that, include: A set of ECG samples is selected from the ECG signal database and input into the detection network to obtain the R peak position of the ECG samples; The time interval (RR interval) between the R peak and its adjacent R peak is calculated based on the R peak position. The RR interval before and after any R peak forms a two-channel RR interval wave. If the fluctuation range of the two-channel RR interval wave is greater than or equal to the threshold, it is marked as "1" to indicate that atrial fibrillation has occurred. If the fluctuation range of the two-channel RR interval wave is less than the threshold, it is marked as "0" to indicate that atrial fibrillation has not occurred. The true label of the two-channel RR interval wave is obtained. The two-channel RR interval waveforms are input into a lightweight atrial fibrillation detection model. Conventional convolutional and depthwise separable convolutional units are used alternately to extract features from the two-channel RR interval waveforms, resulting in feature maps of the two-channel RR interval waveforms. Specifically, the conventional convolutional and depthwise separable convolutional units utilize a lightweight attention module in the second layer to enhance the feature extraction capability of the lightweight atrial fibrillation detection model. The lightweight atrial fibrillation detection model exhibits a symmetrical structure, with four convolutional units on one side of each level. Each convolutional unit includes a one-dimensional convolution, a batch normalization layer, and a leaky rectified linear unit. The detection results were obtained by processing the feature map of the two-channel RR interval wave using the sigmoid activation function. The detection results are fitted and trained with the true labels of the two-channel RR interval waves to obtain a lightweight atrial fibrillation detection model with the minimum loss function value.

2. The atrial fibrillation detection and early warning method according to claim 1, characterized in that, The lightweight attention module is implemented using the CBAM attention mechanism.

3. The atrial fibrillation detection and early warning method according to claim 1, characterized in that, The calculation formula for the depthwise separable convolutional unit is as follows: ， in, The size of the convolution kernel. and These represent the number of input and output channels, respectively.

4. An electrocardiogram (ECG) data relay device, characterized in that, include: The housing contains a main controller module, a Bluetooth module, an OTG module, a storage module, a cellular network and WIFI module, and a power supply module. The outer surface of the housing is provided with a button module, an LCD module, and an LED module in sequence. The main controller module is connected to the Bluetooth module, OTG module, storage module, cellular network and WIFI module, power supply module, button module, LCD module, and LED module respectively, and is used to control the interaction between the various modules. The main controller module includes a lightweight detection unit, which stores a computer program. When the computer program is executed by the main controller module, it implements the steps of the atrial fibrillation detection and early warning method as described in any one of claims 1 to 3, which facilitates atrial fibrillation detection of the electrocardiogram signal in an offline state. The Bluetooth module and the OTG module are used to receive the electrocardiogram (ECG) signals sent by the ECG acquisition device and transmit them to the main controller module; The storage module is used to store the electrocardiogram signal; The cellular network and WIFI module are used to send the electrocardiogram signal to a remote server for atrial fibrillation detection when the network is online. The button module is used to send control signals to the main controller module; The LCD module is used to display the electrocardiogram waveform after the atrial fibrillation detection; The LED module is used for alarming abnormalities in atrial fibrillation detection of electrocardiogram signals; The power supply module is used to supply power to the ECG data relay device.

5. The electrocardiogram data relay device according to claim 4, characterized in that, The Bluetooth module includes an NFC unit, which is used to obtain the Bluetooth MAC address of the ECG acquisition device through NFC identification, pair the ECG data relay device and the ECG acquisition device via Bluetooth, and transmit ECG signals.

6. The electrocardiogram data relay device according to claim 4, characterized in that, Also includes: A SIM card slot is located on the housing of the ECG data relay device and is used to insert a SIM card to transmit ECG signals via a cellular network.

7. The electrocardiogram data relay device according to claim 4, characterized in that, The button module includes a first button and a second button. The first button is used to control the power on / off of the ECG data relay device, and the second button is used to turn the LCD display on / off.

8. The electrocardiogram data relay device according to claim 4, characterized in that, The power supply module is powered by a lithium battery and has an additional charging circuit to charge the battery.

9. The electrocardiogram data relay device according to claim 4, characterized in that, The LCD module can also display the battery level information of the ECG data relay device.

Images