Wireless wearable electrocardiogram monitoring system with self-calibration and motion artifact suppression function

Abstract

The application provides a wireless wearable electrocardio monitoring system with self-calibration and motion artifact suppression function, comprising a flexible sensing unit (100), an edge computing unit (200) and a cloud / terminal analysis platform (300), characterized in that the flexible sensing unit (100) comprises a flexible bioelectrode assembly (110) and a flexible signal processing circuit (120), the flexible signal processing circuit (120) is integrated with a signal amplification circuit (122) and an alternating current impedance excitation source (121); the flexible bioelectrode assembly (110) contains at least one special working electrode, and the working electrode assembly is connected to an electrocardio signal collection channel and an impedance detection channel simultaneously, so as to solve the technical problems of signal drift with impedance, serious motion artifact interference and high power consumption in the prior art.

Description

Technical Field

[0001] This application relates to the field of medical testing equipment technology, and in particular to a wireless wearable electrocardiogram monitoring system with self-calibration and motion artifact suppression functions, which is suitable for high-precision, long-term dynamic electrocardiogram monitoring. Background Technology

[0002] While existing wireless wearable ECG monitoring systems have addressed wearing comfort and basic wireless transmission issues through flexible sensing units and a two-level wireless transmission architecture (NFC + Bluetooth), they still suffer from the following significant drawbacks in practical applications: 1. Skin-electrode contact impedance drift over time: After the flexible electrode comes into contact with the skin, the contact impedance will change significantly within a few hours due to changes in sweat secretion and stratum corneum hydration, leading to signal baseline drift or even saturation. Existing systems lack in-situ, real-time impedance monitoring and compensation mechanisms.

[0003] 2. Insufficient ability to suppress motion artifacts: In motion scenarios (such as running and rehabilitation training), the relative micro-movements between the electrodes and the skin generate triboelectric noise (motion artifacts) that is much greater than that of the ECG signal. Existing systems rely solely on analog filtering with fixed parameters (such as high-pass filtering), which makes it difficult to recover effective QRS complexes from severely contaminated signals, leading to an increased false alarm rate.

[0004] 3. Power consumption and data processing bottlenecks: Although the NFC module in the flexible sensing unit does not require a battery, it often still needs to be powered by a button battery in order to support continuous transmission of multiple channels, which limits the system's ability to monitor for ultra-long periods (such as more than 7 days); and all raw data is transmitted to the terminal for processing, which puts pressure on Bluetooth bandwidth and terminal computing power. Summary of the Invention

[0005] The purpose of this application is to provide a wireless wearable ECG monitoring system with self-calibration and motion artifact suppression functions, so as to solve the technical problems of signal drift with impedance, severe motion artifact interference and excessive power consumption in the prior art.

[0006] To achieve the above objectives, embodiments of this application provide a system that incorporates the following structural improvements based on the existing flexible sensing unit and wireless transmission unit: A wireless wearable ECG monitoring system with self-calibration and motion artifact suppression functions includes a flexible sensing unit 100, an edge computing unit 200, and a cloud / terminal analysis platform 300. The flexible sensing unit 100 includes a flexible bioelectrode assembly 110 and a flexible signal processing circuit 120. The flexible signal processing circuit 120 integrates a signal amplification circuit 122 and an AC impedance excitation source 121. The flexible bioelectrode assembly 110 includes at least one dedicated working electrode, which is simultaneously connected to the ECG signal acquisition path and the impedance detection path.

[0007] The edge computing unit 200 includes a first wireless transceiver circuit 210, an analog-to-digital converter circuit 220, an adaptive filtering coprocessor 230, a modulation and demodulation circuit, and a second wireless transceiver circuit 240. The AC impedance excitation source 121 is configured to output a detection signal of a preset frequency to the flexible bioelectrode assembly 110, and its output terminal is connected to at least one driving electrode in the flexible electrode array; the adaptive filtering coprocessor (230) has a built-in motion artifact recognition algorithm and impedance compensation algorithm, and is configured to calculate the skin-electrode contact impedance based on the response signal generated by the detection signal, and generate a gain control command based on the contact impedance to dynamically adjust the amplification factor of the signal amplification circuit 122.

[0008] Furthermore: The signal amplification circuit in the flexible signal processing circuit 120 uses a programmable gain instrumentation amplifier such as AD8233 or MAX30001, and its gain pin is connected to the GPIO port of the adaptive filter coprocessor 230 to accept digital control.

[0009] Furthermore: The adaptive filtering coprocessor 230 is also connected to a triaxial accelerometer 231; the adaptive filtering coprocessor 230 is configured to execute the variable step size minimum mean square (LMS) algorithm, using the motion signal collected by the triaxial accelerometer 231 as the reference input and the electrocardiogram (ECG) signal as the main input, to update the filtering coefficients in real time and perform adaptive filtering on the ECG digital signal output by the analog-to-digital converter 220 to suppress motion artifacts.

[0010] Furthermore: The step size update factor μ(n) of the variable step size LMS algorithm is positively correlated with the magnitude of the vector sum of the acceleration signals |a(n)|, and the mathematical expression for this relationship is: μ(n)=μmin+(μmax−μmin)⋅∣∣a∣∣max−∣∣a∣∣min∣a(n)∣∣−∣∣a∣∣min; Where μmin and μmax are the preset minimum and maximum step size factors, respectively, |a(n)| is the magnitude of the acceleration vector sum at the current moment, and |a|min and |a|max are the lower and upper limits of the dynamic range of the acceleration magnitude, respectively.

[0011] Furthermore: The edge computing unit 200 also includes an energy harvesting module 250, which is selected from one or a combination of piezoelectric energy harvesters, electromagnetic energy harvesters, or photovoltaic energy harvesters. The output of the energy harvesting module 250 is connected to the power management bus of the edge computing unit 200 for powering the system or charging the energy storage elements.

[0012] Furthermore: The signal amplification circuit 122 is a programmable gain instrumentation amplifier (PGA); the adaptive filter coprocessor 230 sends digital control commands to the PGA via a wireless link to increase the gain to a first gain level when the contact impedance is greater than a first preset threshold, and decrease the gain to a second gain level when the contact impedance is less than a second preset threshold.

[0013] Furthermore: The system is configured to operate in time-division duplex mode: in the first time window, the AC impedance excitation source 121 operates to perform impedance detection; in the second time window, the flexible bioelectrode assembly 110 operates to perform electrocardiogram signal acquisition; the switching frequency of the time windows is higher than 1 kHz.

[0014] Furthermore: The edge computing unit 200 is also configured to perform R-wave peak detection and RR interval calculation on the filtered ECG digital signal. The third wireless transceiver circuit 240 only sends the calculated RR interval sequence and heart rate variability (HRV) parameters to the external terminal, instead of sending the original full-bandwidth waveform data.

[0015] This application has the following beneficial effects: 1. Signal Quality Self-Calibration: By introducing an AC impedance excitation and detection circuit, the system can obtain the contact impedance value of the skin-electrode in real time. When the impedance increases, the system automatically increases the gain of the amplifier circuit or adjusts the cutoff frequency of the high-pass filter to offset the signal attenuation caused by poor contact, realizing end-to-end adaptive calibration and significantly reducing signal drift during long-term monitoring.

[0016] 2. Robust Motion Artifact Suppression: Utilizing the adaptive filter built into the edge computing unit, combined with acceleration data, it can further filter out motion-related low-frequency noise in the digital domain after the analog domain. Compared to traditional fixed filtering, this scheme can maintain a QRS wave detection accuracy of over 90% even under drastic motion, significantly reducing false positive alarms.

[0017] 3. Intelligent power consumption management: By performing preliminary signal quality assessment and feature extraction at the edge (such as extracting only the R-wave peak value and time interval), it is not necessary to transmit all the original waveform data, which greatly reduces the duty cycle of Bluetooth transmission; together with the energy harvesting module, the system can achieve continuous monitoring around the clock or even for several weeks without frequent battery replacements.

[0018] 4. System architecture innovation: The simple "wireless transmission unit" is upgraded to a smart gateway with "edge computing capabilities", which reduces the computing burden of terminal devices and is more in line with the technological development trend of the Internet of Things (IoMT). Attached Figure Description

[0019] Figure 1 This is a system block diagram provided for Embodiment 1 of this application. Detailed Implementation

[0020] The present application will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0021] Example 1: like Figure 1 As shown, the system in this embodiment includes a flexible sensing unit 100 and an edge computing unit 200.

[0022] The flexible sensing unit 100 includes a flexible bioelectrode assembly 110 and a flexible signal processing circuit 120. Unlike existing technologies, in this embodiment, the flexible signal processing circuit 120 not only includes traditional amplification and filtering circuits, but also integrates an AC impedance excitation source 121. This excitation source is controlled by the edge computing unit 200 and outputs a weak sine wave signal of a specific frequency (e.g., 20kHz) to the driving electrode.

[0023] In this embodiment, the AC impedance excitation source 121 is crucial. The edge computing unit 200 sends a command to the flexible sensing unit 100: "Start measuring impedance," and then injects a signal. The AC impedance excitation source 121 then begins to operate, outputting a voltage with a very small peak-to-peak value (e.g., 50mV), which flows into the human tissue through the driving electrode. The working electrode acquires this signal, which has already changed due to passing through the skin and tissue. Impedance is calculated: the circuit measures the "voltage change / current magnitude" to obtain the impedance value Z. Ohm's law applies: Z = V / I. Feedback adjustment: if the measured Z is large (e.g., due to dry skin or sweating causing poor contact), the system automatically increases the amplifier gain; if Z is small (good contact), the gain is maintained or decreased. Simultaneously, the signal acquired by the working electrode in the flexible bioelectrode assembly 110 enters two paths: one is the conventional signal amplification circuit 122 (which can be a programmable gain amplifier PGA), and the other is the impedance detection circuit 123.

[0024] Edge computing unit 200 (which can be designed as a wrist-worn or chest-mounted device) includes: Second wireless transceiver circuit 210 (NFC module): used to receive analog electrocardiogram signals and impedance response signals from flexible sensing unit 100.

[0025] Analog-to-digital converter (ADC) 220: Converts received analog signals into digital signals.

[0026] Adaptive Filtering Coprocessor 230: This is the core of the system. It runs two key algorithms: Impedance compensation algorithm: Analyze the amplitude and phase changes of the 20kHz excitation signal to calculate the current skin-electrode contact impedance Zskin. If Zskin exceeds a threshold (e.g., 500kΩ), the coprocessor 230 sends a control command to the flexible sensing unit 100 via a wireless link to adjust the PGA gain (e.g., from 100x to 200x) to ensure stable output signal amplitude.

[0027] Motion artifact suppression algorithm: The built-in triaxial accelerometer 231 is used to acquire the motion signal a(t), which is then fed into a variable step size LMS filter as a reference input. The filter continuously adjusts the weight coefficients W according to the error signal e(t) (i.e., the difference between the output signal and the desired signal), thereby separating the clean ECG signal s′(t) from the noisy ECG signal s(t)+n(t).

[0028] The third wireless transceiver circuit 240 (BLE 5.0 module): sends only processed digital features (such as RR intervals, heart rate variability HRV data) or compressed waveform data packets to the smartphone or cloud server 300.

[0029] Example 2: Compared to Example 1, this embodiment adds specific details regarding power consumption optimization: In this embodiment, the edge computing unit 200 also includes an energy harvesting module 250 composed of micro photovoltaic cells or piezoelectric thin films. When the human body is stationary or under sufficient sunlight, the energy harvesting module 250 charges the lithium battery. In addition, when the coprocessor 230 detects that the human body is stationary and there are no arrhythmic events, it automatically reduces the sampling rate and transmission frequency, entering a "low-power monitoring mode", thereby greatly extending the battery life.

[0030] Example 3: Compared with Example 1 or Implementation 2, this embodiment adds specific details regarding the optimization of the electrode structure: To ensure that impedance detection and ECG acquisition do not interfere with each other, the flexible electrode array employs time-division multiplexing or multi-frequency multiplexing. For example, in terms of time, 99% of the time is used to acquire ECG signals, and 1% of the time is used to inject a 20kHz excitation signal for impedance detection. Due to the extremely short time interval, the user is unaware of the data and it does not affect the continuity of ECG data.

[0031] Example 4: Compared with any of the above examples, this embodiment and Example 1 achieve dynamic gain control to prevent signal saturation or attenuation.

[0032] When a person exercises, their skin sweats, and the electrodes move slightly. This causes drastic changes in the skin-electrode contact impedance. When the impedance increases, the signal becomes weak (to the point of being invisible); when the impedance decreases (for example, due to good conductivity from sweating), the signal suddenly increases (to the point that the amplifier is overloaded, causing waveform clipping and distortion).

[0033] To address the above issues and achieve dynamic gain control to prevent signal saturation or attenuation, the signal amplification circuit 122 is a programmable gain instrumentation amplifier (PGA). The adaptive filter coprocessor 230 sends digital control commands to the programmable gain instrumentation amplifier via a wireless link to increase the gain to a first gain level when the contact impedance is greater than a first preset threshold, and to decrease the gain to a second gain level when the contact impedance is less than a second preset threshold.

[0034] In other words, this embodiment makes the measured impedance value adjustable. 1. If the impedance is too high, the amplification factor (gain) will be automatically increased to bring back the weak signal.

[0035] 2. If the impedance is too low, the amplification factor will be automatically reduced to prevent the signal from being too strong and damaging the equipment.

[0036] In summary, this embodiment achieves: 1. Constant signal amplitude: Whether you are sitting still or running, the amplitude of the ECG waveform transmitted to the back end remains basically the same.

[0037] 2. Improve signal-to-noise ratio (SNR): Targeted amplification when the signal is weak, rather than using a fixed gain throughout, can significantly improve the recognition rate of weak signals.

[0038] 3. Protection circuit: to prevent strong signals from burning out subsequent circuits or causing ADC (analog-to-digital converter) overload.

[0039] Example 5: This embodiment can significantly eliminate crosstalk between impedance detection and ECG measurement, enabling "measurement and sampling simultaneously".

[0040] To measure impedance, an alternating current signal must be injected into the body. This signal is very strong. If you want to measure the electrocardiogram (ECG) at the same time as the electrical charge is injected, the ECG (microvolt level) will be completely overwhelmed by this strong current (volt level).

[0041] To address the aforementioned crosstalk, this embodiment configures the system to operate in time-division duplex mode: In the first time window, the AC impedance excitation source (121) operates for impedance detection; in the second time window, the flexible bioelectrode assembly (110) operates for electrocardiogram (ECG) signal acquisition. The switching frequency of the time windows is higher than 1 kHz, effectively dividing time into extremely small fragments (>1000 times / second). In other words, in the first time zone (a very short time), ECG acquisition is turned off, impedance detection is turned on, and current is applied to measure impedance. In the second time zone (a slightly longer time), impedance detection is turned off, ECG acquisition is turned on, and the system focuses on measuring the heartbeat. Because the switching speed is extremely fast (>1 kHz), the human heartbeat (1-2 Hz) cannot perceive this switching, but from the circuit's perspective, the two are strictly separated.

[0042] The technical effect of this embodiment is: 1. Pure signal: The ECG pathway is completely unaffected by impedance excitation signals, and the measured ECG baseline is stable without spikes.

[0043] 2. Good real-time performance: It does not require stopping to measure impedance and then stopping to measure ECG, but can monitor continuously.

[0044] 3. Saves hardware: Only one set of electrodes is used to accomplish two tasks, eliminating the need for additional dedicated impedance detection electrodes.

[0045] Example 6: In this embodiment, the edge computing unit 200 is also configured to perform R-wave peak detection and RR interval calculation on the filtered ECG digital signal. The third wireless transceiver circuit 240 only sends the calculated RR interval sequence and heart rate variability (HRV) parameters to the external terminal, instead of sending the original full-bandwidth waveform data. This embodiment can significantly reduce power consumption; transmitting a few bytes of data is much more energy-efficient than transmitting a few KB of waveform data, and the battery life is also significantly extended. At the same time, it improves privacy protection. The cloud and mobile APP cannot see what your specific ECG looks like; they can only see the heart rate variability parameters, fundamentally avoiding the risk of biometric data leakage. Furthermore, this embodiment reduces the computational burden: the mobile phone and cloud do not need to process massive amounts of raw data; they only need to analyze simple parameters, resulting in faster response times.

Claims

1. A wireless wearable ECG monitoring system with self-calibration and motion artifact suppression functions, comprising a flexible sensing unit (100), an edge computing unit (200), and a cloud / terminal analysis platform (300), characterized in that: The flexible sensing unit (100) includes a flexible bioelectrode assembly (110) and a flexible signal processing circuit (120). The flexible signal processing circuit (120) integrates a signal amplification circuit (122) and an AC impedance excitation source (121). The flexible bioelectrode assembly (110) includes at least one dedicated working electrode, which is simultaneously connected to the electrocardiogram signal acquisition path and the impedance detection path. The edge computing unit (200) includes a first wireless transceiver circuit (210), an analog-to-digital converter circuit (220), an adaptive filtering coprocessor (230), a modulation and demodulation circuit, and a second wireless transceiver circuit (240). The AC impedance excitation source (121) is configured to output a detection signal of a preset frequency to the flexible bioelectrode assembly (110), and its output terminal is connected to at least one driving electrode in the flexible electrode array. The adaptive filtering coprocessor (230) has a built-in motion artifact recognition algorithm and impedance compensation algorithm. It is configured to calculate the skin-electrode contact impedance based on the response signal generated by the detection signal, and generate a gain control command based on the contact impedance to dynamically adjust the amplification factor of the signal amplification circuit (122).

2. The system according to claim 1, characterized in that: The signal amplification circuit in the flexible signal processing circuit (120) adopts a programmable gain instrumentation amplifier, whose gain pin is connected to the GPIO port of the adaptive filter coprocessor (230) and accepts digital control.

3. The system according to claim 1, characterized in that: The adaptive filtering coprocessor (230) is also connected to a triaxial accelerometer (231); the adaptive filtering coprocessor (230) is configured to execute the variable step size minimum mean square (LMS) algorithm, using the motion signal collected by the triaxial accelerometer (231) as the reference input and the electrocardiogram (ECG) signal as the main input, to update the filtering coefficients in real time and perform adaptive filtering on the ECG digital signal output by the analog-to-digital converter (220) to suppress motion artifacts.

4. The system according to claim 3, characterized in that: The step size update factor μ(n) of the variable step size LMS algorithm is positively correlated with the magnitude of the vector sum of the acceleration signals |a(n)|, and the mathematical expression for this relationship is as follows: μ(n)=μmin+(μmax−μmin)*∣∣a∣∣max−∣∣a∣∣min∣a(n)∣∣−∣∣a∣∣min Where μmin and μmax are the preset minimum and maximum step size factors, respectively, |a(n)| is the magnitude of the acceleration vector sum at the current moment, and |a|min and |a|max are the lower and upper limits of the dynamic range of the acceleration magnitude, respectively.

5. The system according to claim 1, characterized in that: The edge computing unit (200) also includes an energy harvesting module (250), which is selected from one or a combination of piezoelectric energy harvesters, electromagnetic energy harvesters or photovoltaic energy harvesters. The output of the energy harvesting module (250) is connected to the power management bus of the edge computing unit (200) for powering the system or charging the energy storage element.

6. The system according to claim 1, characterized in that: The signal amplification circuit (122) is a programmable gain instrumentation amplifier (PGA); the adaptive filter coprocessor (230) sends digital control commands to the programmable gain instrumentation amplifier via a wireless link to increase the gain to a first gain level when the contact impedance is greater than a first preset threshold, and to decrease the gain to a second gain level when the contact impedance is less than a second preset threshold.

7. The system according to claim 1, characterized in that: The system is configured to operate in time-division duplex mode: in the first time window, the AC impedance excitation source (121) operates to perform impedance detection; in the second time window, the flexible bioelectrode assembly (110) operates to perform electrocardiogram signal acquisition; the switching frequency of the time windows is higher than 1 kHz.

8. The system according to claim 1, characterized in that: The edge computing unit (200) is also configured to perform R-wave peak detection and RR interval calculation on the filtered ECG digital signal. The third wireless transceiver circuit (240) only sends the calculated RR interval sequence and heart rate variability (HRV) parameters to the external terminal, instead of sending the original full-bandwidth waveform data.

Images