Wearable device and method of detecting atrial fibrillation

Abstract

A wearable device and a method of detecting atrial fibrillation. The method includes: receiving a photoplethysmogram (PPG) signal from a first sensor of the wearable device; detecting a heart rate from the PPG signal based on window power spectrum analysis of the PPG signal; and detecting atrial fibrillation based on the heart rate.

Description

Technical Field

[0001] The methods, apparatuses, and devices according to the present disclosure relate to medical monitoring methods and devices, and more particularly to wearable devices and methods for detecting atrial fibrillation. Background Art

[0002] Atrial fibrillation (AF) is a heart disease that causes rapid and irregular heartbeats in the atria and can lead to serious health risks such as blood clots and stroke. Detection algorithms based on various biosignals such as electrocardiogram (ECG) or photoplethysmogram (PPG) signals have been developed to detect irregular heart rhythms such as AF or atrial flutter (AFL). ECG-based detection has high accuracy, but requires multiple electrodes to be attached to the patient to measure the ECG signal, and multiple electrodes need to be in contact with human skin, and are therefore difficult to monitor continuously in wearable devices. The advantage of PPG-based detection is continuous monitoring, but it is susceptible to motion artifacts. Summary of the Invention

[0003] One aspect provides a wearable device and method for atrial fibrillation detection, which improves the accuracy of atrial fibrillation diagnosis by continuously monitoring atrial fibrillation based on photoplethysmogram (PPG) and electrocardiogram (ECG) signals.

[0004] According to an aspect of one or more embodiments, there is provided a method for detecting atrial fibrillation, the method comprising: receiving a photoplethysmogram (PPG) signal from a first sensor of a wearable device; detecting a heart rate from the PPG signal based on windowed power spectrum analysis of the PPG signal; and detecting atrial fibrillation based on the heart rate.

[0005] According to another aspect of one or more embodiments, there is provided a method for detecting atrial fibrillation, the method comprising: acquiring a photoplethysmogram (PPG) signal from a first sensor of a wearable device; acquiring a motion detection signal from a second sensor of the wearable device; determining whether there is motion exceeding a reference value based on the motion detection signal; detecting a heart rate from the PPG signal based on a peak-to-peak interval (PPI) when the motion does not exceed the reference value, and detecting a heart rate from the PPG signal based on a power spectrum analysis of the PPG signal in a frequency domain when the motion exceeds the reference value; and detecting atrial fibrillation based on the heart rate.

[0006] According to yet another aspect of one or more embodiments, a wearable device is provided, comprising: a first sensor configured to sense a pulse wave of a user and generate a photoplethysmogram (PPG) signal based on the pulse wave; a second sensor configured to sense motion of the user and generate a motion detection signal; a memory storing program code; and at least one processor configured to access the memory to execute the program code, wherein the program code causes at least one of the at least one processor to: detect a heart rate based on a peak-to-peak interval (PPI) of the PPG signal when the second sensor does not sense motion, and detect the heart rate based on a power spectrum analysis of the PPG signal in a frequency domain when the second sensor senses motion, and wherein the program code causes at least one of the at least one processor to detect atrial fibrillation based on the heart rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Various embodiments will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings, in which:

[0008] Figure 1 is a block diagram illustrating an electronic device according to an embodiment;

[0009] Figure 2 is a block diagram illustrating a first sensor according to an embodiment;

[0010] Figure 3A and Figure 3B An example in which the electronic device according to some embodiments is implemented as a wearable device is shown;

[0011] Figure 4A and Figure 4B is a signal waveform diagram showing an example of a photoplethysmogram (PPG) signal and an electrocardiogram (ECG) signal;

[0012] Figure 5 is a flow chart illustrating a method of detecting atrial fibrillation according to an embodiment;

[0013] Figure 6 is a flow chart illustrating windowed power spectrum analysis according to an embodiment;

[0014] Figure 7 and Figure 8 A method for detecting multiple heart rates by window power spectrum analysis according to an embodiment is shown;

[0015] Figure 9 shows a PPG signal in the time domain and a window power spectrum in the frequency domain when there is motion of the user;

[0016] Figure 10 is a flow chart illustrating windowed power spectrum analysis according to an embodiment;

[0017] Figure 11 shows the removal of motion artifacts according to an embodiment;

[0018] Figure 12 is a flow chart illustrating windowed power spectrum analysis according to an embodiment;

[0019] Figure 13 shows a finite state machine (FSM) according to an embodiment;

[0020] Figure 14 is a flow chart illustrating a method of detecting atrial fibrillation according to an embodiment;

[0021] Figure 15 is a flowchart illustrating a method for detecting a heart rate based on a peak-to-peak interval (PPI) characteristic of a PPG signal according to an embodiment;

[0022] Figure 16 is a flowchart illustrating an operating method of an electronic device according to an embodiment; and

[0023] Figure 17A and Figure 17B A biosignal monitoring system according to some embodiments is shown. DETAILED DESCRIPTION

[0024] Hereinafter, various embodiments will be described in detail with reference to the accompanying drawings.

[0025] Figure 1 An electronic device according to an embodiment is shown.

[0026] refer to Figure 1 , the electronic device 100 may include a processor 110, a sensor module 120, an input / output device 130, a communication module 140, a memory 150, a storage device 160, and a power module 170. The electronic device 100 is not limited thereto and may further include various components.

[0027] The electronic device 100 may be a wearable device for monitoring a user's biological signals. The user may wear the electronic device 100 on a part of the body, such as an arm, leg, or neck. The electronic device 100 may sense the user's biological signals and monitor the user's health status by using sensors provided in the sensor module 120 (e.g., the first sensor 121, the second sensor 122, and the third sensor 123).

[0028] The sensor module 120 may include a plurality of sensors, for example, a first sensor 121, a second sensor 122, and a third sensor 123. The first sensor 121, the second sensor 122, and the third sensor 123 may sense different biosignals.

[0029] The first sensor 121 may be a photoplethysmogram (PPG) sensor. The PPG sensor may generate a PPG signal by measuring the user's pulse wave. The PPG sensor may measure the user's pulse wave based on the principle that the amount of light absorbed by the user's skin varies according to the heartbeat when light is irradiated onto the user's skin.

[0030] Reference Figure 2 , the first sensor 121 may include a light emitting module 21, a light receiving module 22 and an analog front end (AFE) 23. The light emitting module 21 may include one or more light emitting elements. For example, the light emitting module 21 may include a light emitting diode (LED). The light receiving module 22 may include one or more photodiodes (PD). For example, the light receiving module 22 may include an optical sensor including a photodetector. Figure 2 In FIG. 2 , the light emitting module 21 is shown as including two LEDs, and the light receiving module 22 is shown as including three PDs, but the embodiment is not limited thereto, and the number of light emitting elements and the number of PDs may each vary.

[0031] In response to a measurement request from the processor 110, the AFE 23 may drive the light emitting module 21 and transmit a sensing signal received from the light receiving module 22 to the processor 110. The AFE 23 may include a driving circuit 24 and a sensing circuit 25.

[0032] The driving circuit 24 may provide a driving signal to the light emitting element. For example, the driving circuit 24 may provide a driving current to the LED. The driving circuit 24 may include a metal oxide silicon field effect transistor (MOSFET) and a digital-to-analog converter configured to control the current.

[0033] The sensing circuit 25 can convert the reflected light measured by the PD into a PPG signal. The sensing circuit 25 may include an amplifier, a filter, and an analog-to-digital converter. For example, the amplifier may be implemented as a transimpedance amplifier.

[0034] The sensing circuit 25 can convert the measured reflected light into a voltage signal using an amplifier and filter the converted voltage signal using a low-pass filter. For example, the low-pass filter can block frequency components greater than 5 Hz (hertz). The analog-to-digital converter can convert the filtered signal into a digital signal, such as a PPG signal.

[0035] The first sensor 121 can continuously measure the user's pulse wave (e.g., continuously as long as the user does not interrupt) and generate a PPG signal. The light-emitting module 21 and the light-receiving module 22 of the first sensor 121 can be in contact with the user's skin, so that the first sensor 121 can measure the user's pulse wave. For example, the first sensor 121 can continuously measure the pulse wave without the user's perception.

[0036] The first sensor 121 may transmit the PPG signal to the processor 110. For example, the transmission of the PPG signal to the processor 110 may be performed based on an interface based on a serial peripheral interface (SPI). However, embodiments are not limited thereto, and an inter-integrated circuit (I2C), an improved inter-integrated circuit (I3C), a mobile industry processor interface (MIPI), a universal asynchronous receiver / transmitter (UART), an embedded display port (eDP), a low voltage differential signaling (LVDS), a universal serial interface (USI), an ultra-path interface (UPI), and / or enhanced step-down differential signaling (eRVDS) may be used between the first sensor 121 and the processor 110.

[0037] Continue to refer Figure 1 , the second sensor 122 may be an electrocardiogram (ECG) sensor. The ECG sensor may generate an ECG signal by measuring the user's ECG. The ECG sensor may include a plurality of electrodes (or a plurality of pads) (e.g., Figure 3A 35a and Figure 3B 35b) and AFE in.

[0038] An electrical signal may be applied to the user's skin through at least one electrode of the plurality of electrodes, and an ECG signal representing electrical activity in the heart may be output through at least one other electrode of the plurality of electrodes.

[0039] At least one of the plurality of electrodes may be arranged to contact the user's skin, and at least one other of the plurality of electrodes may come into contact with the user's skin when the user intentionally touches the electrode with a part of the body (e.g., a finger). Thus, the second sensor 122 may measure an ECG in response to a user's request (e.g., a touch), or may measure an ECG in response to a request from the processor 110. The AFE may generate an ECG signal by amplifying and analog-to-digital converting the measured electrical signal.

[0040] The second sensor 122 may transmit the ECG signal to the processor 110. For example, the transmission of the ECG signal to the processor 110 may be performed based on an SPI-based interface. However, the embodiment is not limited thereto, and one of the various high-speed serial interface (HSSI) methods described above may be applied between the second sensor 122 and the processor 110.

[0041] The third sensor 123 may be a motion detection sensor. For example, the third sensor 123 may be implemented as an inertial measurement unit (IMU). The IMU may include an accelerometer and a gyroscope. The IMU may also include a geomagnetic sensor. The third sensor 123 may generate an IMU signal, such as a 3-axis accelerometer signal, by measuring user motion (hereinafter referred to as motion), such as a posture change, a change speed of position movement, or a displacement amount. The IMU signal may be referred to as a motion detection signal. The third sensor 123 may continuously measure user motion and generate an IMU signal.

[0042] The third sensor 123 may transmit the IMU signal to the processor 110. For example, the transmission of the IMU signal to the processor 110 may be performed based on an SPI-based interface. However, the embodiment is not limited thereto, and one of the various HSSI methods described above may be applied between the third sensor 123 and the processor 110.

[0043] In an embodiment, the sensor module 120 may further include other biometric sensors. For example, the sensor module 120 may further include a sensor configured to measure the user's bioimpedance and a sensor configured to sense the state or change of sweat, blood, urine, and / or iris. For example, the sensor module 120 may further include a galvanic skin response (GSR) sensor, an electrodermal activity (EDA) sensor, a ballistocardiogram (BCG) sensor, a sweat sensor for sensing water content or dehydration, an iris sensor, and / or a body temperature sensor.

[0044] The processor 110 may control the overall operation of the electronic device 100 and control components such as the sensor module 120, the input / output device 130, the communication module 140, the memory 150, the storage device 160, and the power module 170. In an embodiment, the processor 110 may include a microcontroller unit (MCU). However, the embodiment is not limited thereto, and the processor 110 may include a processing circuit system such as a central processing unit (CPU) or a microprocessing unit (MPU). In some embodiments, the processor 110 may include at least one processor, and at least one of the at least one processor may perform the following operations with respect to Figures 5 and 6 、 Figure 10 、 Figure 12 and Figures 14 to 16In some embodiments, the processor 110 may include multiple processors 110. In some embodiments, the multiple processors 110 may execute the following operations: Figures 5 and 6 、 Figure 10 、 Figure 12 and Figures 14 to 16 In some embodiments, each of the plurality of processors 110 may execute the following operations: Figures 5 and 6 、 Figure 10 、 Figure 12 and Figures 14 to 16 In some embodiments, the processor 110 may include multiple cores, each of which is configured to execute the following operations: Figures 5 and 6 、 Figure 10 、 Figure 12 and Figures 14 to 16 One or more operations described.

[0045] The processor 110 may process biosignals received from the sensor module 120 , such as a PPG signal, an ECG signal, and an IMU signal, and monitor a user's health status based on the signals.

[0046] In an embodiment, the processor 110 may detect atrial fibrillation based on a PPG signal. The processor 110 may monitor in real time whether atrial fibrillation has occurred based on the PPG signal continuously received from the first sensor 121. The processor 110 may detect a heart rate based on the PPG signal and detect atrial fibrillation based on the detected heart rate. For example, the processor 110 may perform a stochastic analysis on a plurality of continuously detected heart rates and determine whether atrial fibrillation has occurred based on the analysis results. For example, Shannon entropy, sample entropy, and / or root mean square difference (RMSSD) may be used as the stochastic analysis.

[0047] When atrial fibrillation is detected, that is, when it is determined that atrial fibrillation has occurred, the processor 110 may generate an event indicating that atrial fibrillation has occurred and output an ECG measurement request signal (or alarm) to the user. The processor 110 may control the second sensor 122 to measure ECG.

[0048] The processor 110 can transmit the ECG signal received from the second sensor 122 to an external device via the communication module 140. In an embodiment, the processor 110 can perform signal processing (sampling, compression, etc.) on the ECG signal to generate ECG data and transmit the ECG data to the external device. For example, the ECG signal (or ECG data) can be transmitted to a smartphone and then transmitted to a medical staff server via the smartphone. Medical staff can diagnose whether atrial fibrillation occurs based on the ECG signal.

[0049] In an embodiment, processor 110 may detect heart rate from the PPG signal based on windowed power spectrum analysis of the PPG signal, and may detect atrial fibrillation based on the detected heart rate. In an embodiment, when motion is determined to have occurred, processor 110 may remove motion artifacts from the motion PPG signal based on the IMU signal. This analysis of the PPG signal and removal of motion artifacts may be performed in the frequency domain. This analysis and operation will be described in detail later.

[0050] In an embodiment, the processor 110 may determine whether motion has occurred based on the IMU signal. For example, when it is determined that there is no motion (e.g., when the magnitude of motion based on the IMU signal is less than or equal to a reference value), the processor 110 may detect the heart rate based on the peak-to-peak interval (PPI) characteristic of the PPG signal, and when it is determined that there is motion (e.g., when the magnitude of motion based on the IMU signal exceeds the reference value), the processor 110 may detect the heart rate from the PPG signal based on windowed power spectrum analysis of the PPG signal as described above.

[0051] In an embodiment, the processor 110 may select a highly reliable heart rate from the detected heart rates by using a finite state machine (FSM). In other words, the processor 110 may discard unreliable heart rates. The processor 110 may detect atrial fibrillation based on the highly reliable heart rate.

[0052] The input / output device 130 may include various devices configured to receive user input and may include various devices configured to provide information, notifications, etc. to the user. The input / output device 130 may include a display 131 and an audio module 132. The input / output device 130 may also include devices such as a vibration module or input keys.

[0053] The display 131 can display various information under the control of the processor 110. For example, the display 131 can display the user's biometric information, such as heart rate or blood oxygen saturation. The display 131 can display atrial fibrillation detection information (whether atrial fibrillation has occurred), arrhythmia information (the presence of arrhythmia and / or the type of arrhythmia), or suspected disease information. The display 131 can display: information requesting user action, for example, information requesting a finger to touch the ECG measurement electrodes for ECG measurement; or recommendation information about hospital visits. For example, when atrial fibrillation is detected based on the PPG signal, the display 131 can display event information indicating that atrial fibrillation has occurred, and display information requesting the user to measure ECG and touch the ECG measurement electrodes with a finger to measure ECG.

[0054] The display 131 may include at least one of various displays such as a liquid crystal display (LCD), a thin film transistor LCD (TFT-LCD), an organic light emitting diode (OLED) display, a light emitting diode (LED) display, an active matrix organic LED (AMOLED) display, a micro LED display, a mini LED display, a flexible display, and a three-dimensional display. In an embodiment, the display 131 may be implemented in the form of a touch screen. In an embodiment, the display 131 may be implemented as a fixed display or a flexible display.

[0055] The audio module 132 can output sound and, for example, can include at least one of an audio codec, a microphone (MIC), a receiver, an earphone output, and a speaker. Based on the acquired biometric information and / or suspected disease information, the audio module 132 can output information related to the user's physical condition, information related to abnormal signs of the user's health condition, or additional information as an audio signal. For example, if atrial fibrillation is detected, the audio module 132 can output an alarm indicating that atrial fibrillation has occurred.

[0056] The communication module 140 can communicate with external devices. In an embodiment, the communication module 140 may include a Bluetooth module. However, the embodiment is not limited thereto, and for example, in some embodiments, the communication module 140 may include a communication interface that can access a wireless local area network (WLAN), such as wireless fidelity (Wi-fi), wireless personal area network (WPAN), wireless universal serial bus (wireless USB), Internet of Things (Zigbee), near field communication (NFC) and radio frequency identification (RFID), or the communication interface can access a mobile communication network, such as third generation (3G), fourth generation (4G) or long term evolution (LTE). In an embodiment, the communication module 140 may also include a communication interface that can access a wired local area network.

[0057] The communication module 140 can send information related to the user's physical condition, information related to abnormal signs of the user's health status, and / or additional information to an external electronic device (e.g., a user's smartphone). The communication module 140 can send the measured biosignals, such as ECG data, to the external electronic device. The external electronic device can send the biosignals to a medical server. In an embodiment, the communication module 140 can access a network (e.g., an access point or a mobile communication network) and send the ECG data directly to the medical server.

[0058] The memory 150 may be implemented as a volatile memory such as a dynamic random access memory (DRAM) and a static RAM (SRAM), or a non-volatile resistive memory such as a phase change RAM (PRAM) and a resistive RAM (ReRAM). In an embodiment, the memory 150 may be integrated into the processor 110. In an embodiment, the memory 150 may include a plurality of memories 150. The plurality of memories 150 may be volatile memories or non-volatile memories, or different combinations thereof.

[0059] An operating program, program code, or application executed by the processor 110 may be loaded into the memory 150 and executed. For example, a program including program code that implements the aforementioned functions of the processor 110 when executed may be loaded into the memory 150 and executed by the processor 110.

[0060] The memory 150 may store data to be processed by the processor 110 or data generated by the processor 110. For example, the memory 150 may temporarily store biometric signal measurement records (e.g., the number of measurements or measurement time), user's biometric information, and suspected disease information.

[0061] The storage device 160 may be implemented as a non-volatile memory device, such as a NAND flash memory or a resistive memory, and, for example, the storage device 160 may be configured as a memory card (MMC card, eMMC card, SD or micro SD card) or the like. The storage device 160 may store data generated by the processor 110. The storage device 160 may store biometric signal measurement records (e.g., the number of measurements or the measurement time), the user's biometric information, and suspected disease information. In an embodiment, the storage device 160 may store measured biometric signals, such as ECG signals.

[0062] The power module 170 may include a battery, a charging circuit, and a power management unit (PMU). In an embodiment, the PMU may be integrated into the processor 110. The power module 170 may generate and provide power used by the electronic device 100 based on power provided from the battery or an external power source. The power module 170 may charge the battery based on the external power source. The PMU may manage the power of the components. For example, the PMU may provide power to a component and determine the level (e.g., voltage level) or operating frequency of the power provided to the component based on the operating state of the electronic device 100 or the operating state of each component. The PMU may block power.

[0063] Figure 3A and Figure 3B An example in which the electronic device according to the embodiment is implemented as a wearable device is shown. Figure 3A1 shows a first surface (eg, front surface) of the electronic device 100, Figure 3B A second surface (eg, rear surface) of the electronic device 100 is shown.

[0064] refer to Figure 3A and Figure 3B The electronic device 100 may be, for example, a watch-type wearable device wearable on a user's wrist or a wearable device wearable on other parts of the human body (e.g., the head, forearm, thigh, or other parts of the human body for measuring ECG).

[0065] The electronic device 100 may include a display 33, input buttons 34, a plurality of electrodes (e.g., a first electrode 35a and a second electrode 35b, etc.), at least one light-emitting element 36, and at least one light-receiving element 37, and these components may be located in a housing 31 forming the appearance of the electronic device 100. A band 32 that assists in wearing the electronic device 100 on the user's body may be connected to the housing 31. Figure 1 Some of the components described, for example, the processor 110 , the communication module 140 , the third sensor 123 , the audio module 132 , the memory 150 , the storage device 160 , and the power module 170 may be disposed inside the housing 31 .

[0066] The first electrode 35a and the second electrode 35b may be electrodes for ECG measurement. The first electrode 35a may be provided on the first surface, and the second electrode 35b may be provided on the second surface. At least one light emitting element 36 and at least one light receiving element 37 may be provided on the second surface. At least one light emitting element 36 and at least one light receiving element 37 may be elements configured to measure the user's pulse wave and may be included in the first sensor 121 ( Figure 1 (in Chinese).

[0067] When the electronic device 100 is worn on a user's body, the second surface can be in contact with the user's skin. Therefore, the at least one light-emitting element 36 and the at least one light-receiving element 37 can always be in contact with the user's skin. Therefore, the first sensor 121 can continuously measure the pulse wave.

[0068] When the electronic device 100 is worn, the second electrode 35b is always in contact with the user's skin. When the electronic device 100 is worn, the first electrode 35a does not contact the user's skin unless there is an intentional action by the user. Therefore, when the user touches the first electrode 35a with a part of the body (e.g., a finger), the second sensor 122 can measure ECG based on the electrical signal through the first electrode 35a and the second electrode 35b.

[0069] Figure 4A and Figure 4Bis a signal waveform diagram showing a PPG signal and an ECG signal. Figure 4A shows the PPG signal and ECG signal when there is no movement, and Figure 4B The figure shows the PPG signal and ECG signal when there is motion. The horizontal axis represents time, and the vertical axis represents the magnitude of the signal voltage.

[0070] Reference Figure 4A The ECG signal can include repeatedly generated P waves (P), Q peaks (Q), R peaks (R), S peaks (S), and T waves (T). The ECG signal includes a prominent feature known as the QRS complex, which represents the primary pumping contraction of the heart. The heart rate algorithm uses the R peaks (R) in the ECG signal to measure the time between the pulses of the pulsation. The duration between R peaks is called the RR interval (RRI).

[0071] The PPG signal is used to represent the periodicity of the signal waveform and includes quasi-periodic pulses with peaks (P) and valleys, which are used to estimate heart rate. The duration between the peaks (P) of two adjacent pulses is called the PP interval (PPI), which can be used as an indicator of heart rate.

[0072] refer to Figure 4B When there is motion, the PPG signal may be distorted by motion artifacts. Therefore, it is difficult to detect the peak (P) in the PPG signal. On the other hand, the ECG signal has less distortion due to motion artifacts than the PPG signal, and the R peak (R) can be detected in a specific form.

[0073] Figure 5 is a flowchart illustrating a method of detecting atrial fibrillation according to an embodiment, and Figure 6 The embodiment according to the present invention is shown in more detail. Figure 5 Flowchart of the window power spectrum analysis in operation S130 of the method. Figure 5 The method may include operations S110, S130, and S150, and Figure 6 The flowchart shown in may include operations S210 to S240. Figures 5 and 6 Operations S110, S130, S150, and S210 to S240 of the embodiment may be performed by Figure 1 and the electronic device 100 of FIG. 3 executes and refers to Figure 1 describe.

[0074] The electronic device 100 may acquire a PPG signal (S110). For example, the processor 110 may control the first sensor 121 to measure the user's pulse wave to generate a PPG signal, and may provide the PPG signal to the processor 110. The PPG signal may be acquired continuously.

[0075] The electronic device 100 may detect the heart rate based on the window power spectrum analysis of the PPG signal (S130). Figure 6 , the processor 110 may generate a PPG segment (e.g., PPG window data) by sampling the PPG signal based on a time window (S210). The time window may be set to a specific time. The processor 110 may filter the PPG segment (S220). For example, the processor 110 may filter the PPG segment by using a bandpass filter. As a non-limiting example, the bandpass filter may pass components in a frequency band of about 0.5 Hz to about 4 Hz. Thus, high-frequency components, such as noise components, may be blocked in the PPG segment. In an embodiment, the first sensor 121 (see Figure 2 ) performs operations S210 and S220 in the AFE 23 of the AFE 23. For example, the sensing circuit 25 of the AFE 23 may generate a PPG segment by sampling the PPG signal based on a time window (S210) and block high-frequency components of the PPG segment by using a bandpass filter. The filtered PPG segment may be sent to the processor 110.

[0076] The processor 110 may convert the PPG segment, for example, the filtered PPG segment, into a power spectrum in the frequency domain, for example, a first power spectrum (S230). In an embodiment, the processor 110 may convert the PPG segment in the time domain into a first power spectrum in the frequency domain by using a fast Fourier transform (FFT).

[0077] The processor 110 may detect the heart rate based on the first power spectrum (S240). For example, the processor 110 may detect a peak in the first power spectrum and calculate the heart rate based on the frequency corresponding to the peak. For example, when the first power spectrum has a peak at 1.2 Hz, the heart rate may be calculated as 72 beats per minute (BPM) by multiplying 1.2 Hz by 60.

[0078] Continue to refer Figure 5 , the electronic device 100 may detect atrial fibrillation based on the detected heart rate (S150). In some embodiments, in operation S130, multiple heart rates corresponding to multiple PPG segments may be detected, and the electronic device 100 may detect atrial fibrillation based on the multiple heart rates. In an embodiment, the processor 110 may determine whether atrial fibrillation occurs through random analysis.

[0079] Figure 7 and Figure 8 A method for detecting multiple heart rates by window power spectrum analysis according to an embodiment is shown. In an embodiment, Figure 7 and Figure 8 The method may be performed by the processor 110 ( Figure 1 (in) execution.

[0080] refer to Figure 7 , processor 110( Figure 1 The PPG signal S can be analyzed based on a time window TW set to a specific time. PPG The PPG segment may include the sampled PPG signal. The time window TW may be along the PPG signal S at specific time intervals. PPG The PPG signal S is shifted to generate a plurality of PPG segments, for example, a first PPG segment FG1, a second PPG segment FG2 and a third PPG segment FG3. PPG It can also be described as sliding in the time window TW. For example, the time window TW can be set to 8 seconds, and the time window TW can be along the PPG signal S PPG Slip and can be used to monitor the PPG signal S at 1 second intervals PPG Take samples. Figure 7 As shown, the first PPG segment FG1 may include, for example, a PPG signal S PPG The second PPG segment FG2 may include, for example, a PPG signal S PPG The four peaks (peak 2 to peak 5), and so on.

[0081] The processor 110 can calculate multiple heart rates by performing power spectrum analysis on each of a plurality of PPG segments, such as the first PPG segment FG1, the second PPG segment FG2, and the third PPG segment FG3. Through power spectrum analysis, an average heart rate over a time window can be calculated, rather than an instantaneous heart rate. For example, a first heart rate HR1 corresponding to the first PPG segment FG1 can be calculated as 74 BPM, a second heart rate HR2 corresponding to the second PPG segment FG2 can be calculated as 76 BPM, and a third heart rate HR3 corresponding to the third PPG segment FG3 can be calculated as 80 BPM.

[0082] refer to Figure 8 , a signal of various frequency components including high-frequency components and low-frequency components of the PPG segment FG can be provided. The high-frequency components and low-frequency components can be removed from the PPG segment by bandpass filtering (BPF). In other words, noise in the high-frequency components and low-frequency components can be removed.

[0083] The filtered PPG segment FG can be converted into a power spectrum in the frequency domain using an FFT. The PPG signal resembles a sine wave, so when the PPG segment FG is converted into a power spectrum, the frequency with the maximum power (e.g., the peak of the power spectrum), i.e., the peak frequency Fpeak, can represent the heart rate. The heart rate can be calculated by multiplying the peak frequency Fpeak by 60.

[0084] Figure 9 The PPG signal in the time domain and the power spectrum in the frequency domain when there is motion of the user are shown.

[0085] When the user moves (and therefore the electronic device 100 being worn by the user moves), the PPG signal may be distorted due to motion artifacts, and peak detection may be difficult. For example, as shown in the figure, errors may occur in some peaks P1 and P2. In other words, peaks P1 and P2 may form at the wrong time, or motion artifacts may be inaccurately identified as peaks. As a result, it may be difficult to detect heart rate based on the PPG signal, and the reliability of the detected heart rate may be reduced.

[0086] However, according to embodiments, when the PPG segment FG is converted into a power spectrum in the frequency domain using a PPG window based on windowed power spectrum analysis, the power spectrum may have high power at several frequencies due to motion artifacts. However, the peak frequency with the maximum power can be easily detected and converted into the heart rate. Therefore, the method of detecting heart rate using windowed power spectrum analysis in the frequency domain can more easily detect heart rate than the method of detecting heart rate based on the PPI characteristics of the PPG signal in the time domain, and can improve heart rate reliability.

[0087] Figure 10 FIG3 is a flowchart illustrating window power spectrum analysis according to an embodiment. The flowchart may include operations S310 to S350. Figure 10 Operations S310 to S350 of the embodiment may be performed by Figure 1 and the electronic device 100 of FIG. 3 executes and refers to Figure 1 describe.

[0088] refer to Figure 10 , the processor 110 may generate a PPG segment by sampling the PPG signal based on a time window (S310). The time window may be set to a specific time. The processor 110 may filter the PPG segment (S320). The processor 110 may convert the PPG segment into a power spectrum in the frequency domain, such as a first power spectrum (S330). In an embodiment, operations S310, S320, and S330 are respectively Figure 6Operations S210, S220, and S230 are the same, and for the sake of brevity, their repeated description is omitted. In an embodiment, in S310 or S320, the processor 110 may filter the PPG signal or PPG segment using a finite impulse response (FIR) filter, a Wiener filter, and / or a Gaussian filter to remove noise and / or motion artifacts. In an embodiment, operation S320 may be performed after operation S330. For example, the processor 110 may convert the PPG segment into a first power spectrum and then perform filtering on the first power spectrum in the frequency domain.

[0089] The processor 110 may generate a second power spectrum by eliminating motion artifacts from the first power spectrum (S340). In an embodiment, the processor 110 may convert the IMU signal received from the third sensor 123 into a third power spectrum in the frequency domain and eliminate the third power spectrum from the first power spectrum. Thus, a second power spectrum with motion artifacts eliminated may be generated.

[0090] The processor 110 may detect the heart rate based on the second power spectrum (S350). The processor 110 may calculate the heart rate based on the peak frequency having the maximum power.

[0091] Figure 11 The removal of motion artifacts according to an embodiment is shown.

[0092] Motion artifacts may have frequency components that overlap with the PPG signal, and it may be difficult to eliminate these frequency components by filtering, such as filtering using a bandpass filter in operation S220 or S320. Therefore, when the PPG signal is distorted by motion artifacts, the power spectrum of the PPG signal (the power spectrum of the PPG segment), such as the first power spectrum SP1, may include multiple peaks, such as a first peak P1, a second peak P2, and a third peak P3. At least two of the first peak P1, the second peak P2, and the third peak P3 may be peaks caused by motion artifacts.

[0093] To identify peaks caused by motion artifacts, the IMU signal received from the third sensor 123 may be converted to the frequency domain using an FFT. For example, when receiving a PPG signal, the processor 110 may simultaneously receive the IMU signal and generate IMU segments by sampling the IMU signal based on a time window. In embodiments, the time points and time periods for sampling the IMU signal based on the time window may be the same as the time points and time periods for sampling the PPG signal based on the time window.

[0094] The processor 110 may generate a power spectrum of the IMU signal in the frequency domain by converting the IMU segments into the frequency domain using an FFT.

[0095] As shown in the figure, the power spectrum SP3 of the IMU signal may include a first peak P1 and a third peak P3. Therefore, it can be seen that the first peak P1 and the third peak P3 in the first power spectrum SP1 are peaks caused by motion artifacts.

[0096] The processor 110 may eliminate the third power spectrum SP3 from the first power spectrum SP1. In other words, the processor 110 may remove the first peak P1 and the third peak P3 in the third power spectrum SP3 from the first power spectrum SP1 to generate the second power spectrum SP2. Therefore, the first peak P1 and the third peak P3 may be eliminated from the first peak P1, the second peak P2, and the third peak P3, and the second power spectrum SP2 may include the second peak P2. The heart rate may be calculated based on the frequency of the second peak P2.

[0097] Figure 12 FIG4 is a flowchart illustrating window power spectrum analysis according to an embodiment. The flowchart may include operations S410 to S460. Figure 12 Operations S410 to S460 of the embodiment may be performed by Figure 1 and the electronic device 100 of FIG. 3 executes and refers to Figure 1 describe.

[0098] refer to Figure 12 , the processor 110 may generate a PPG segment by sampling the PPG signal based on a time window (S410). For example, the time window may be set to a specific time. The processor 110 may filter the PPG segment (S420). The processor 110 may convert the PPG segment into a first power spectrum in the frequency domain (S430), and remove motion artifacts from the first power spectrum to generate a second power spectrum (S440). The processor 110 may detect the heart rate based on the second power spectrum (S450). Operations S410 to S450 are similar to Figure 10 Operations S310 to S350 are the same, and their repeated descriptions are omitted for brevity.

[0099] The processor 110 may selectively output the heart rate (S460) by using a finite state machine (FSM). For example, the processor 110 may use the FSM to determine whether the heart rate is stably detected, and when it is determined that the heart rate is stably detected, the processor 110 may output the heart rate to detect atrial fibrillation. When it is determined that the heart rate is not stably detected, the processor 110 may discard the heart rate.

[0100] Figure 13 An FSM according to an embodiment is shown.

[0101] refer to Figure 13, FSM is an algorithm for determining the reliability of the detected heart rate, and may include a first state ST1, a second state ST2, a third state ST3, and a fourth state ST4. In an embodiment, the FSM may be generated by the processor 110 ( Figure 1 (in) run.

[0102] The first state ST1 may be a stable state, the second state ST2 may be a recovery state, the third state ST3 may be a pulse state, and the fourth state ST4 may be an atrial fibrillation (AF) state. The first state ST1 indicates a state in which the heart rate is stably detected, that is, a state in which the PPG signal is stably measured. The second state ST2 indicates a state in which the PPG signal is not stably measured and an incorrect heart rate is estimated to be detected. The third state ST3 indicates a state in which there is a temporary abnormality in the heartbeat and it is necessary to check whether a PPG signal is measured from the user. The fourth state ST4 indicates a state in which it is determined that, despite the temporary abnormality in the heartbeat, a PPG signal is measured from the user and the heart rate is stably detected.

[0103] Regardless of whether the heart rate is abnormal, the first state ST1 and the fourth state ST4 both represent states where the heart rate is stably detected.

[0104] The second state ST2 and the third state ST3 both represent states where the heart rate is not stably detected. Therefore, the heart rates in the first state ST1 and the fourth state ST4 have high reliability and can therefore be output for detecting atrial fibrillation. The heart rates in the second state ST2 and the third state ST3 have low reliability, and therefore, when the heart rates in the second state ST2 and the third state ST3 are used to detect atrial fibrillation, the accuracy of the atrial fibrillation detection result may be low. Therefore, the heart rates in the second state ST2 and the third state ST3 may not be output and may be discarded.

[0105] When it is determined that the heart rate varies excessively within a short period of time, the processor 110 may determine that the PPG signal is not measured in a stable state and convert the current state in which the heart rate is detected to the second state ST2 or the third state ST3. The processor 110 may determine the presence or absence of a main peak in the second power spectrum based on a crest factor indicating how distinct the peaks are, which may be a signal generated by a heartbeat, and when it is determined that the main peak does not exist, the processor 110 may make a transition to a corresponding state depending on the previous state. Such state transitions may be performed by an FSM and are described below with reference to Figure 13 Detailed description.

[0106] exist Figure 13In the figure, HRN represents normal heart rate variability, and CF(stable), CF(recovery), and CF(af) represent the crest factors of the first state ST1, second state ST2, and fourth state ST4, respectively. An exclamation point (!) before a term indicates that a condition is not met. For example, "!HRN" means that the heart rate variability is not normal, that is, abnormal. cn(recovery) represents the number of times the recovery condition, such as the condition for the second state ST2, third state ST3, and fourth state ST4 to transition to the first state ST1, is met. Nrecovery, Npulse, and Nafrehab represent the number of times cn(recovery) is required for the second state ST2, third state ST3, and fourth state ST4 to transition to the first state ST1. cn(af) represents the number of times the third state ST3 meets the condition for transitioning to the fourth state ST4, and Naf represents the number of times cn(af) is required for the third state ST3 to transition to the fourth state ST4.

[0107] Based on heart rate variability ΔHR, normal heart rate variability HRN can be determined. According to Equation 1, the absolute value of the difference between the current (i-th) heart rate HR(i) and the previous (i-1-th) heart rate HR(i-1) can be calculated as heart rate variability ΔHR. Here, the previous heart rate HR(i-1) is the heart rate output for detecting atrial fibrillation.

[0108] Equation 1

[0109] ΔHR=|HR(i)-HR(i-1)|

[0110] When the heart rate variability ΔHR is less than the threshold set for the heart rate variability, it can be determined that the heart rate variability is not large (normal heart rate variability HRN), and when the heart rate variability ΔHR is equal to or greater than the threshold, it can be determined that the heart rate variability is large (abnormal heart rate variability! HRN).

[0111] For example, when the threshold is set to 5, the current heart rate HR(i) is 75BPM and the previous heart rate HR(i-1) is 78BPM, the heart rate variability (ΔHR) is 3, and therefore it can be determined that the heart rate variability is not large (normal heart rate variability HRN).

[0112] The crest factor CF can be calculated according to Equation 2.

[0113] Equation 2

[0114] CF(i)=Xpeak(i) / Xrms(i)

[0115] Here, CF(i) represents the crest factor in the second spectrum of the PPG segment in which the current heart rate HR(i) is detected, Xpeak(i) represents the power value of the frequency with the highest power in the second spectrum, and Xrms(i) represents the average power of all frequencies in the second spectrum. In other words, the crest factor indicates how much greater the power at a specific frequency is than the average power in all frequency bins of the frequency range.

[0116] A first threshold value THstable, a second threshold value THrecovery, and a third threshold value THaf for determining the crest factors of the first state, the second state, and the fourth state may be set. When CF(i) is greater than the first threshold value THstable, CF(i) may be estimated as the crest factor CF(stable) in the first state, when CF(i) is less than or equal to the first threshold value THstable and greater than the second threshold value THrecovery, CF(i) may be estimated as the crest factor CF(recovery) in the second state, and when CF(i) is less than or equal to the second threshold value THrecovery and greater than the third threshold value THaf, CF(i) may be estimated as the crest factor CF(af) in the fourth state.

[0117] In the first state ST1, a peak that will become the main peak of the heart rate signal exists in the second power spectrum, and heart rate variability is normal. For example, when the previous state (the state in which the previous heart rate (HR(i-1)) is detected) is the first state ST1 and the current heart rate HR(i) is detected, heart rate variability is normal (HRN), and when CF(i) is the crest factor CF(stable) in the first state, the current state can be maintained in the first state ST1 (matn1).

[0118] In the second state ST2, there is no main peak in the second power spectrum, and the heart rate variability may be normal. In this case, the detected heart rate may be noise caused by motion artifacts rather than a heart rate signal.

[0119] When the previous state is the first state ST1, CF(i) is not the crest factor CF(stable) in the first state (!CF(stable)), and when the heart rate variability is normal (HRN), the first state ST1 may be converted to the second state ST2 (TRS1). The current heart rate HR(i) may not be output and may be discarded.

[0120] The current heart rate HR(i) can be recalculated based on a PPG segment generated after the PPG segment of the current heart rate HR(i) is detected. Heart rate variability and the current crest factor CF(i) can be calculated based on the recalculated current heart rate HR(i). When the heart rate variability is normal (HRN) and the current crest factor CF(i) is the crest factor CF(recovery) in the second state, the first recovery condition CF(recovery) & HRN is satisfied, for transitioning the second state ST2 to the first state ST1. Cn(recovery) can increase (cn(recovery)++). When the first recovery condition CF(recovery) & HRN is satisfied several times consecutively and cn(recovery) reaches Nrecovery, the second state ST2 can transition to the first state ST1 (TRS2). When the current state is converted to the first state ST1 or the first recovery condition CF(recovery)&HRN is not satisfied (!(CF(recovery)&HRN)), cn(recovery) may be initialized to 0 (cn(recovery)=0).

[0121] When the previous state is the first state ST1, if the heart rate variability is abnormal (!HRN), the first state ST1 may be converted to the third state ST3 (TRS3). The current heart rate HR(i) may not be output and may be discarded.

[0122] Then, when the heart rate variability is normal (HRN) and the current crest factor CF(i) is not the crest factor CF(af) in the fourth state (!CF(af)), the second recovery condition HRN & !CF(af) for transitioning from the third state ST3 to the first state ST1 is satisfied. When the second recovery condition HRN & !CF(af) is satisfied several times consecutively and cn(recovery) reaches Npulse, the third state ST3 can be transitioned to the first state ST1 (TRS4).

[0123] When heart rate variability is abnormal (!HRN) and the current crest factor CF(i) is the crest factor CF(af) in the fourth state, the first judgment condition for transitioning from the third state ST3 to the fourth state ST4, !HRN & CF(af), is met. If the first judgment condition, !HRN & CF(af), is met several times consecutively and cn(af) reaches Naf, the third state ST3 can be transitioned to the fourth state ST4 (TRS5).

[0124] The fourth state ST4 is a state in which heart rate variability is large and CF(af) is continuously satisfied, and the fourth state ST4 is determined to be a state in which a cardiac physiological abnormality exists but the heart rate is stably detected. Then, when the heart rate variability is normal (HRN) and the current crest factor CF(i) is the crest factor CF(af) in the fourth state, the third recovery condition HRN & CF(af) for the fourth state ST4 to be converted to the first state ST1 is satisfied. When the third recovery condition HRN & CF(af) is satisfied several times in a row and cn(recovery) reaches Nafrehab, the fourth state ST4 can be converted to the first state ST1 (TRS6). In other words, when the current crest factor CF(i) maintains the crest factor CF(af) in the fourth state and the heart rate variability is stable, the fourth state ST4 can be converted to the first state ST1 (TRS6).

[0125] In the first state ST1 and the fourth state ST4, the current heart rate HR(i) is output. In the second state ST2 and the third state ST3, it may be determined that the heart rate is not stably detected, and the detected current heart rate HR(i) may be discarded and redetected (recalculated). Therefore, when it is determined that the heart rate has high reliability, in other words, when a PPG signal in a stable state is measured, the processor 110 can detect atrial fibrillation based on the detected heart rate, thereby improving the accuracy of atrial fibrillation detection.

[0126] Figure 14 is a flow chart illustrating a method of detecting atrial fibrillation according to an embodiment. Figure 15 is a flow chart illustrating a method for detecting a heart rate based on the PPI characteristic of a PPG signal. Figure 14 The flowchart may include operations S510 to S550, and Figure 15 The flowchart may include operations S610 to S640. Figure 14 and Figure 15 Operations S510 to S550 and operations S610 to S640 of the embodiment may be performed by Figure 1 and the electronic device 100 of FIG. 3 executes and refers to Figure 1 describe.

[0127] refer to Figure 14 , the electronic device 100 may acquire a PPG signal and a motion detection signal (S510). For example, the processor 110 may receive a PPG signal from the first sensor 121 and an IMU signal from the third sensor 123.

[0128] The electronic device 100 may check whether motion is detected (S520). For example, the processor 110 may determine whether user motion is present based on the IMU signal from the third sensor 123. For example, the processor 110 may calculate motion artifacts of the IMU signal and determine that motion is present when the motion artifacts exceed a reference value, and determine that motion is absent when the motion artifacts are less than or equal to the reference value. The reference value may be pre-set.

[0129] When no motion is detected (ie, when the motion artifact is less than or equal to the reference value) (operation S520, No), the electronic device 100 may detect the heart rate based on the PPI characteristics of the PPG signal (S530). Figure 15 , processor 110 (or Figure 2 The sensing circuit 25) can filter the PPG signal to remove noise (S610). For example, as a noise removal filter, an infinite impulse response (IIR) filter, a moving average filter, and a heart rate (HR) range filter can be used.

[0130] The processor 110 may detect the PPI of the filtered PPG signal (S620). Figure 4A As shown, the duration between peaks P of two adjacent pulses among a plurality of pulses of the filtered PPG signal can be detected as PPI.

[0131] Processor 110 may remove outliers from the detected plurality of PPIs (S630). Outliers are more likely to be caused by noise. Therefore, outliers may be removed to improve the accuracy of the heart rate. For example, when the difference with other adjacent PPIs is equal to or greater than a set removal threshold, the corresponding PPI may be determined as an outlier and removed.

[0132] The processor 110 may calculate a heart rate based on the detected plurality of PPIs (S640).

[0133] Continue to refer Figure 14 When motion is detected (ie, when the motion artifact is greater than the reference value) (operation S520, yes), the electronic device 100 may detect the heart rate based on the window power spectrum analysis of the PPG signal (S540). Figures 6 to 13 As described, the electronic device 100 can detect heart rate based on windowed power spectrum analysis of a PPG signal. The PPG signal can be sampled based on a time window, generated into PPG segments, and converted into a power spectrum in the frequency domain. Motion artifacts can be removed from the power spectrum, and a highly reliable heart rate can be output based on the FSM.

[0134] The electronic device 100 may detect atrial fibrillation based on the detected heart rate (S550). For example, the processor 110 may check whether atrial fibrillation occurs based on the detected heart rate.

[0135] Figure 16 FIG. 1 is a flowchart illustrating an operating method of an electronic device according to an embodiment of the present invention. The method may include operations S710 to S760. Figure 16 Operations S710 to S760 of the method shown may be performed by Figure 1 electronic equipment to perform.

[0136] refer to Figure 16 , the electronic device 100 may acquire a PPG signal (S710) and detect a heart rate based on a windowed power spectrum analysis of the PPG signal (S720). The electronic device 100 may detect atrial fibrillation based on the detected heart rate (S730). Operations S710, S720, and S730 are similar to Figure 5 Operations S110, S130, and S150 are the same, and therefore, their repeated descriptions are omitted for brevity.

[0137] When atrial fibrillation is detected, the electronic device 100 may notify the user of the abnormality (S740). For example, the electronic device 100 may output event information notifying the user of the occurrence of the abnormality (ie, detection of atrial fibrillation) through the display 131.

[0138] The electronic device 100 may acquire an ECG signal (S750). The user may check the abnormality notification and touch the electrode of the second sensor 122 with a body part (e.g., a finger) to measure the ECG. The electronic device 100 may measure the ECG and generate an ECG signal. Thus, the electronic device 100 may acquire an ECG signal in response to detecting atrial fibrillation.

[0139] The electronic device 100 may output the ECG signal to an external device (S760). As described above, whether atrial fibrillation occurs can be monitored at all times based on PPG. However, when medical personnel analyze the ECG signal, a more accurate diagnosis of atrial fibrillation can be performed. Therefore, the electronic device 100 may output the ECG signal to an external device to allow medical personnel to diagnose atrial fibrillation based on the ECG signal. In some embodiments, the external device may be, for example, a medical personnel server. In some embodiments, the external device may be a mobile device such as a smart phone, and the ECG signal may be sent to the medical personnel server via the mobile device.

[0140] Figure 17A and Figure 17B A biosignal monitoring system according to some embodiments is shown.

[0141] refer to Figure 17A and Figure 17B , the biosignal monitoring systems 1000a and 1000b may include a biosignal monitoring device 1100 and a data receiving device 1200. The biosignal monitoring system 1000b may further include a server 1300.

[0142] refer to Figure 1 The electronic device 100 described above, such as a wearable device, can be applied to the biosignal monitoring device 1100. Therefore, the above description of the electronic device 100 and its operation can be applied to the biosignal monitoring device 1100, and for the sake of brevity, a repeated description thereof is omitted. The data receiving device 1200 can be an electronic device including a mobile communication interface, such as a smartphone, tablet PC, or mobile communication device. The server 1300 can be, for example, a medical staff server or a cloud server.

[0143] The biosignal monitoring device 1100 can measure PPG signals, ECG signals, and motion detection signals (e.g., IMU signals). The biosignal monitoring device 1100 can detect heart rate through windowed power spectrum analysis of the PPG signal and detect atrial fibrillation based on the heart rate. The biosignal monitoring device 1100 can continuously monitor whether atrial fibrillation has occurred based on the PPG signal. In an embodiment, the biosignal monitoring device 1100 can eliminate motion artifacts from the PPG signal based on the motion detection signal. In an embodiment, when atrial fibrillation is determined to have occurred, the biosignal monitoring device 1100 can output event information notifying the user of the occurrence of an abnormal condition (i.e., detection of atrial fibrillation) through a display and / or a speaker. When the user touches the electrodes with a part of their body to measure the ECG, the biosignal monitoring device 1100 can measure the ECG and send the ECG signal to the data receiving device 1200 through a wired short-range communication interface or a wireless short-range communication interface.

[0144] The data receiving device 1200 may transmit the ECG signal (or signal-processed ECG data) to the server 1300. In an embodiment, the biosignal monitoring device 1100 may include a mobile communication interface and may transmit the ECG signal directly to the server 1300. Medical personnel may detect (diagnose) atrial fibrillation based on the ECG signal transmitted to the server 1300.

[0145] As described above, the electronic device 100 ( Figure 1 ) and biosignal monitoring device 1100 ( Figure 17A and Figure 17B(In) atrial fibrillation can be detected based on the PPG signal. When there is motion, the electronic device 100 and the biosignal monitoring device 1100 can eliminate motion artifacts from the PPG signal by using a window power spectrum method in the frequency domain, can detect the heart rate by window power spectrum analysis, and can detect atrial fibrillation based on a highly reliable heart rate by using FSM. Therefore, the electronic device 100 and the biosignal monitoring device 1100 can continuously monitor whether atrial fibrillation occurs, and can improve the accuracy of atrial fibrillation detection based on the PPG signal. The electronic device 100 and the biosignal monitoring device 1100 can measure ECG when atrial fibrillation is detected based on the PPG signal, and send the ECG signal to an external electronic device or a medical staff server for diagnosis of atrial fibrillation. Therefore, the continuity and accuracy of atrial fibrillation detection can be improved.

[0146] Various embodiments have been described in the drawings and the specification. Although specific terms are used in this specification to describe the various embodiments, these terms are used solely for the purpose of explaining the technical concepts of the present disclosure and are not intended to limit the meaning or scope of the appended claims. Therefore, those skilled in the art will appreciate that various modifications and other equivalent embodiments are possible. Therefore, the true scope of technical protection should be determined by the technical spirit of the appended claims.

[0147] While various embodiments have been particularly shown and described with reference to the accompanying drawings, it will be understood that various changes in form and details may be made therein without departing from the spirit and scope of the appended claims.

Claims

1. A method for detecting atrial fibrillation, the method comprising: receiving a photoplethysmogram (PPG) signal from a first sensor of the wearable device; detecting a heart rate from the PPG signal based on a windowed power spectrum analysis of the PPG signal; as well as Atrial fibrillation is detected based on the heart rate.

2. The method according to claim 1, wherein: Detecting the heart rate includes: generating a PPG segment by sampling the PPG signal based on a time window; Converting the PPG segment into a first power spectrum in the frequency domain; removing motion artifacts from the first power spectrum to generate a second power spectrum; and The heart rate is estimated based on the second power spectrum.

3. The method according to claim 2, wherein: Estimating the heart rate includes converting a frequency having a peak in the second power spectrum into the heart rate.

4. The method according to claim 2, wherein: Eliminating the motion artifacts includes: converting a motion detection signal received from a second sensor of the wearable device into a third power spectrum in the frequency domain; and The third power spectrum is eliminated from the first power spectrum to generate the second power spectrum.

5. The method according to claim 2, further comprising: determining a state in which the heart rate is detected based on a finite state machine; as well as Based on a state indicating that the heart rate is stably calculated, the heart rate is output for detecting the atrial fibrillation.

6. The method according to claim 5, wherein: The finite state machine includes a stable state, a recovery state, a pulse state and an atrial fibrillation state; The state in which the heart rate is detected shifts between the stable state, the recovery state, the pulse state, and the atrial fibrillation state based on the presence or absence of a main peak in the second power spectrum, the crest factor of the main peak, and the heart rate variability, and The heart rate is output in the stable state or the atrial fibrillation state, and the heart rate is discarded in the recovery state or the pulse state.

7. The method according to claim 2, wherein: Generating the PPG segments comprises generating a plurality of PPG segments by sliding the PPG signal in the time window at specific time intervals. wherein the plurality of PPG segments include a first PPG segment and a second PPG segment, and The first PPG segment and the second PPG segment partially overlap in time.

8. The method according to claim 7, wherein: A plurality of heart rates are detected from the plurality of PPG segments, and Wherein, detecting the atrial fibrillation comprises performing a random analysis on the plurality of heart rates.

9. The method of claim 1, further comprising detecting the presence or absence of motion based on a motion detection signal received from a second sensor of the wearable device, in, Detecting the heart rate includes detecting the heart rate from the PPG signal based on a peak-to-peak interval (PPI) characteristic of the PPG signal when there is no motion.

10. The method according to claim 1, further comprising: In response to detecting the atrial fibrillation, acquiring an electrocardiogram (ECG) signal through a third sensor of the wearable device; as well as The ECG signal is sent to an external device for performing an atrial fibrillation diagnosis based on the ECG signal.

11. A method for detecting atrial fibrillation, the method comprising: Acquire a photoplethysmogram (PPG) signal from a first sensor of the wearable device; Acquiring a motion detection signal from a second sensor of the wearable device; determining whether there is motion based on the motion detection signal; When the movement amplitude based on the movement detection signal does not exceed a reference value, it is determined that there is no movement, and the heart rate is detected from the PPG signal based on a peak-to-peak interval PPI of the PPG signal; and when the movement amplitude based on the movement detection signal exceeds the reference value, it is determined that there is the movement, and the heart rate is detected from the PPG signal based on a power spectrum analysis of the PPG signal in the frequency domain; as well as Atrial fibrillation is detected based on the heart rate.

12. The method according to claim 11, further comprising: Based on the detection of the atrial fibrillation, acquiring an electrocardiogram (ECG) signal through a third sensor of the wearable device; as well as The ECG signal is sent to an external device for performing an atrial fibrillation diagnosis based on the ECG signal.

13. The method according to claim 11, wherein: Detecting the heart rate from the PPG signal based on the power spectrum analysis comprises: generating a PPG segment and a motion segment by sampling the PPG signal and the motion detection signal based on a time window; Convert the PPG segment and the motion segment into a first power spectrum and a third power spectrum in the frequency domain, respectively; eliminating the third power spectrum from the first power spectrum to generate a second power spectrum; and The heart rate is estimated based on the second power spectrum.

14. The method according to claim 13, further comprising: identifying a state in which the heart rate is calculated using a finite state machine including a steady state, a recovery state, a pulse state, and an atrial fibrillation state; When the state is the stable state or the atrial fibrillation state, outputting the heart rate for detecting the atrial fibrillation; as well as When the state is the recovery state or the pulse state, the heart rate is not output and is discarded.

15. The method according to claim 14, wherein: Based on the presence or absence of a main peak in the second power spectrum, the crest factor of the main peak, and the heart rate variability, the current state is transferred between the stable state, the recovery state, the pulse state, and the atrial fibrillation state.

16. The method according to claim 11, wherein: Detecting the heart rate includes: A stochastic analysis is performed on a plurality of heart rates detected from a plurality of PPG segments, wherein the plurality of PPG segments temporally overlap with each other.

17. A wearable device, comprising: a first sensor configured to sense a pulse wave of a user and generate a photoplethysmogram (PPG) signal based on the pulse wave; a second sensor configured to sense a motion of the user and generate a motion detection signal; a memory storing program code; as well as at least one processor configured to access the memory to execute the program code, wherein the program code causes at least one of the at least one processor to: detect the heart rate based on a peak-to-peak interval (PPI) of the PPG signal when the second sensor does not sense motion, and detect the heart rate based on a power spectrum analysis of the PPG signal in a frequency domain when the second sensor senses the motion, and The program code causes at least one of the at least one processor to detect atrial fibrillation based on the heart rate.

18. The wearable device according to claim 17, further comprising: a third sensor configured to sense an electrocardiogram (ECG) of the user and generate an electrocardiogram (ECG) signal based on the ECG; monitor; as well as Communication interface, Wherein, based on detecting the atrial fibrillation, the program code causes at least one of the at least one processor to: control the third sensor to sense the ECG and generate the ECG signal, and control the communication interface to send the ECG signal to an external device for atrial fibrillation diagnosis.

19. The wearable device of claim 17, wherein: The program code causes at least one of the at least one processor to generate a PPG segment and a motion segment by sampling the PPG signal and the motion detection signal based on a time window when the second sensor senses motion, The program code causes at least one of the at least one processor to: convert the PPG segment and the motion segment into a first power spectrum and a third power spectrum in the frequency domain, respectively, and generate a second power spectrum by eliminating the third power spectrum from the first power spectrum, and The program code causes at least one of the at least one processor to estimate the heart rate based on the second power spectrum.

20. The wearable device of claim 19, wherein: The program code causes at least one of the at least one processor to identify a state in which the heart rate is calculated using a finite state machine including a steady state, a recovery state, a pulse state, and an atrial fibrillation state, When the state is the stable state or the atrial fibrillation state, the program code causes at least one of the at least one processor to output the heart rate for detecting the atrial fibrillation, and When the state is the recovery state or the pulse state, the program code causes at least one of the at least one processor to not output the heart rate and discard the heart rate.

Images