Intelligent wearable heart rate detection device
By employing a multi-LED, multi-PD probe layout and signal synthesis and enhancement processing, the problem of insufficient accuracy in heart rate detection during exercise in existing smart wearable devices has been solved, achieving high robustness and accuracy in heart rate detection under various environments.
Patent Information
- Application Number
- CN202411675659.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-21
- Publication Date
- 2025-01-10
AI Technical Summary
Existing smart wearable heart rate monitoring devices are not very accurate under strenuous exercise conditions, especially those based on PPG signals, which are easily affected by ambient light, poor tissue blood perfusion, and motion artifacts, resulting in insufficient robustness and accuracy.
By employing a multi-LED, multi-PD probe layout and optimizing the distance and angle between the LEDs and PDs, combined with lock-in amplification technology and signal synthesis enhancement processing, and through synchronous acquisition and demodulation of multi-channel signals, optimization problems are defined to perform signal synthesis enhancement, thereby improving the AC/DC ratio and signal quality of the PPG signal.
Under conditions of strenuous exercise, it significantly improves the accuracy and robustness of heart rate detection, and can reliably obtain accurate heart rate values compared with electrocardiograms.
Smart Images
Figure CN119279548A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to an intelligent wearable heart rate detection device.
[0002] Heart rate is an important indicator of human health. Its value and changes can reflect the health of the heart and the human body. Continuous heart rate monitoring is of great significance in health assessment, exercise guidance and daily health care.
[0003] Smart wearable devices generally use photoplethysmography to detect heart rate. Photoplethysmography is a non-invasive method to measure changes in blood volume in the microvascular bed of the skin based on the optical properties of absorption, scattering and transmission of various components in human tissue under the action of light of a specific wavelength. Based on the Lambert-Beer law, the light attenuation of incident light of the same wavelength is related to the concentration of the absorber. By detecting the intensity of the reflected light, the changes in blood volume in the tissue over time can be continuously monitored. The principle of heart rate detection through photoplethysmography is that the light emitted by the light-emitting diode (LED) of a specific wavelength, after attenuation by human tissue, is detected by the photodetector (PD) and outputs a current corresponding to the light intensity. The voltage after the current-voltage (IV) conversion is collected by the analog-to-digital conversion (AD) module, and finally the signal is processed on the processor.
[0004] The morphology of the photoplethysmogram (PPG) pulse is the result of a complex interaction between the left ventricle and the systemic circulation. It consists of an early main pulse generated by ventricular contraction and various additional pulses caused by pressure pulses reflected from the central arterial tree to the peripheral vasculature. As the main pulse reaches the first reflection site, the junction of the thoracic and abdominal aorta, the arterial diameter is significantly reduced and the arterial elasticity changes, causing the main pulse to be reflected. The main pulse continues to move downward and reaches the second reflection site, which originates from the junction of the abdominal aorta and the common iliac artery. These reflection sites are often called renal and iliac reflection sites and produce second and third reflection waves. The heart rate can be calculated by detecting the distance between the PPG peaks or the frequency of the corresponding peaks in the spectrum.
[0005] Compared with the electrocardiogram, the gold standard for heart rate detection, PPG has the advantages of being non-invasive, portable and low-cost. In particular, it can provide single-arm continuous monitoring, which makes it convenient for people to conduct self-monitoring in their lives. Background Art
[0006] Although there are many smart wearable devices based on vascular volume sensors on the market, such as smart bracelets, watches, rings, etc., which nominally have the function of detecting heart rate, recent studies have found that compared with the gold standard electrocardiogram for heart rate detection, the accuracy of heart rate detection is not easy to guarantee, especially in the case of strenuous exercise, and the measured value is usually underestimated. [1] , the robustness of heart rate detection is not ideal.
[0007] A typical PPG-based wearable heart rate detection device includes a light-emitting diode, a photodetector, an AD detection module, and a processor. Most smart wearable heart rate detection devices on the market use a configuration of one LED and one PD. Currently, some smart watches from brands such as Huawei and Apple also use a configuration of multiple LEDs and multiple PDs.
[0008] The quality of the PPG signal is easily affected by the characteristics of the light-emitting diode and the photodetector, as well as the pressure applied to the PPG probe, which may affect the morphology of the PPG waveform. In addition, ambient light at the photodetector, poor blood perfusion of the surrounding tissue, and motion artifacts are also common sources of error. In an uncontrolled environment, these potential error sources are more frequent and can become a serious obstacle to the reliable use of PPG heart rate detection, especially in continuous monitoring applications. Improving the robustness of detection is an important research direction in PPG detection.
[0009] The paper "Wearable Multichannel Photoplethysmography Framework for HeartRate Monitoring During Intensive Exercise" uses 9 light sensors with a distance of 7mm between adjacent light sensors to simultaneously acquire multiple PPG signals at slightly different positions, and then uses multi-channel singular value decomposition to reduce motion artifacts. However, it does not consider the optimization of the distance between the green light LED and the PD to enhance the maximum AC / DC ratio of the signal. [2] .
[0010] The paper "Design and Validation of a New PPG Module to Acquire High-Quality Physiological Signals for High-Accuracy Biomedical Sensing" studied the optimal LED-PD distance that results in the maximum AC / DC ratio of the PPG signal under different wavelengths of light, and designed, optimized, and fabricated a new photoplethysmographic PPG module to obtain high-quality PPG signals required for high-accuracy biosensing. However, it only uses one photodetector and does not consider the signal measurement when the detector is greatly affected by noise. [3] .
[0011] The paper "Development of a human heart rate monitoring system based on photoplethysmography" uses a three-dimensional acceleration sensor and an adaptive signal algorithm to filter out motion artifacts, but it does not use multiple channels and does not consider the distance between the PD and the LED to improve the light intensity detected by the PD. [4].
[0012] It can be seen that many existing studies on heart rate detection of smart watch PPG signals have tried to use multi-channel methods or add acceleration sensors to try to improve detection robustness, but there is still a lack of designs for multiple LEDs and multiple PDs while paying attention to the distance between LEDs and PDs. Therefore, a reasonable layout is needed to improve the AC / DC ratio of PPG signals based on multiple channels and a solution to provide subsequent signal processing methods based on the corresponding system.
[0013] References
[0014] [1] Pilar M, Maria AC, Luisa MD, et al. Are Activity Wrist-Worn DevicesAccurate for Determining Heart Rate during Intense Exercise? [J].Bioengineering,2023,10(2):254-254.
[0015] [2]HL, HC, HK, et al. Wearable Multichannel PhotoplethysmographyFramework for Heart Rate Monitoring During Intensive Exercise[J]. IEEE SensorsJournal, 2018, 18(7): 2983-2993.
[0016] [3]Hua YK, P. CPC, Long C W.Design and Validation of a New PPGModule to Acquire High-Quality Physiological Signals for High-AccuracyBiomedical Sensing[J]. IEEE Journal of Selected Topics in Quantum Electronics, 2019, 25(1): 1-10.
[0017] [4] Cheng Jie. Development of human heart rate monitoring system based on photoplethysmography[D]. Chongqing University of Technology, 2022.DOI: 10.27753 / d.cnki.gcqgx.2022.000158. Summary of the invention
[0018] Purpose of the Invention
[0019] In view of the shortcomings of existing wearable heart rate detection devices in terms of robustness and accuracy, the present invention intends to integrate the characteristics of some existing solutions and propose a new solution for intelligent wearable heart rate detection equipment, which improves the robustness and accuracy of heart rate detection through multi-channel detection with reasonable sensor layout and signal synthesis enhancement processing.
[0020] Technical Solution
[0021] An intelligent wearable heart rate detection device, which includes LED, PD, IV conversion circuit, AD acquisition circuit, processor and program, characterized in that: (1) an optimized multi-LED, multi-PD probe layout is adopted to make the PPG signal obtain a larger AC-DC ratio, and a corresponding multi-channel IV conversion circuit and signal synchronization acquisition scheme are adopted; (2) the collected multi-channel photoelectric conversion signals are demodulated using phase-locked amplification technology to obtain multi-channel pulse wave signals; (3) an optimization problem is defined for the demodulated multi-channel pulse wave signals, and synthesis enhancement processing is performed on them, and then the heart rate is calculated based on the enhanced signals. Figure 1 shown.
[0022] According to the above-mentioned intelligent wearable heart rate detection device, an optimized multi-LED, multi-PD probe layout is adopted, characterized in that 4 multi-color LED package products and 4 photodetectors are selected, each multi-color LED package integrates green, red and infrared wavelengths of light, the geometric center of the 4 LED package products and the geometric center of the effective photosensitive area of the 4 PDs are located on 2 concentric circles with different radii, the LED devices and the PDs are staggered, and the central angle corresponding to the geometric center of each LED device and the geometric center of the effective photosensitive area of the adjacent PD is maintained at 45°, in order to obtain a better maximum AC / DC ratio of the PPG signal, the green LED is kept closer to the center of the circle relative to the red and infrared LEDs, and the distance between the LED and the PD is appropriate. Figure 2 shown.
[0023] According to the above-mentioned intelligent wearable heart rate detection device, its circuit design should match the probe solution, and is characterized in that the circuit includes an LED driving circuit, an IV conversion circuit, a signal acquisition circuit, etc., and the single-chip microcomputer generates a PWM wave of a specified frequency to control the on and off of the NMOS tube, thereby controlling the flashing frequency of the LED; the heart rate mode can be selected in the host computer to make the four green LEDs flash at the same time at a specified frequency to measure the heart rate, or the blood oxygen mode can be selected to make the four red LEDs and the four infrared LEDs flash alternately at a specified frequency; the IV conversion circuit converts the photocurrent output by the photodetector into a voltage signal of a suitable magnitude, and the voltage signal is collected by the ADC acquisition circuit.
[0024] According to the above-mentioned intelligent wearable heart rate detection device, its data acquisition should match the probe solution, which is characterized in that the original data of 7 channels are synchronously collected, channels 1 to 4 are the voltages after the photocurrents of PD1 to PD4 are converted through IV, and channels 5 to 7 correspond to the PWM waves driving red light, infrared light, and green light, respectively.
[0025] According to the above-mentioned intelligent wearable heart rate detection device, the collected multi-channel photoelectric conversion signals are demodulated using the phase-locked amplification technology, which is characterized in that the 4-channel PPG raw data are multiplied by the PWM wave that drives the green light LED respectively and then filtered and pre-processed to remove high-frequency noise and baseline drift to obtain a 4-channel pulse wave signal.
[0026] According to the above-mentioned intelligent wearable heart rate detection device, an optimization problem is defined for the multi-channel pulse wave signals obtained by demodulation, and synthesis enhancement processing is performed on them. It is characterized in that the objective function of the optimization problem is defined as the sum of the highest spectral peak of the synthetic signal spectrum and the amplitude of its second harmonic and third harmonic, so as to find a set of normalized combination vectors that maximize the objective function, and synthesize and enhance the 4-channel pulse signals obtained by phase-locked amplification.
[0027] Beneficial Effects
[0028] In order to verify the beneficial effects of the present invention, a comparative test was conducted. In the experiment, we selected 4 subjects with good health conditions and compared the device of the present invention with the electrocardiogram, the gold standard for heart rate detection. Figure 3 As shown, the experimental subject wears the device of the present invention on one wrist, and at the same time, the ECG electrode clips are clamped on both hands, and the sampling rate is 10kHz. Each experimental subject measures multiple sets of data, and the pre-processed 4-channel PPG signal, the synthetic enhanced PPG signal and the ECG signal are respectively subjected to zero-filled FFT to make the amplitude spectrum frequency resolution reach 0.01. The heart rate obtained by the PPG signal of each channel, the synthetic signal and the ECG signal are compared according to the heart rate calculated by the spectrum peak.
[0029] In Experiment 1, 10s of the PPG signal measured by the experimental subject was selected for synthesis enhancement, such as Figure 4 As shown in the figure, the time domain waveforms of the 4-channel PPG signals after sequential preprocessing and the synthesized enhanced signals all have high signal quality and good waveforms, which are consistent with the characteristics of the pulse wave waveform. After comparison with the ECG signal, the heart rate measurement results of each channel and the synthesized enhanced signal are accurate.
[0030] In Experiment 2, 10s of the PPG signal measured by the experimental subject was selected for synthesis enhancement, such as Figure 5As shown in the figure, due to the presence of noise, the peaks of the spectrum of channels 1, 2, 3 and 4 do not correspond to the frequency measured by the ECG signal. A set of coefficients [0.6178 0.2627 0.6597 0.3378] are optimized to implement synthesis enhancement. The frequency corresponding to the spectrum peak of the synthetic signal is the same as the heart rate frequency measured by ECG, and the heart rate value obtained is 88.8 beats / minute (i.e. Figure 5 1.48Hz in ).
[0031] In Experiment 3, 10s of the PPG signal measured by the experimental subject was selected for synthesis enhancement, such as Figure 6 As shown in the figure, due to the presence of noise, the spectrum peaks of channel 2, channel 3 and channel 4 do not correspond to the heart rate frequency. A set of coefficients [0.7194 0.3442 0.3251 0.5083] are optimized to implement synthesis enhancement. The frequency corresponding to the spectrum peak of the synthetic signal is the same as the heart rate frequency measured by ECG, and the heart rate value obtained is 102.6 beats / minute (i.e. Figure 6 1.71Hz in ).
[0032] In the above experiments, for different experimental subjects, when the spectral peaks of each channel are the same and when the spectral peaks of the ECG signals in different channels are offset, the device of the present application can stably obtain the accurate average heart rate. It can be seen that the device of the present application has good robustness and accuracy in heart rate measurement. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 This is a logic block diagram of this application.
[0034] Figure 2 This is the probe distribution diagram of this application.
[0035] Figure 3 This is a diagram of the detection experiment scene of this application.
[0036] Figure 4 This is a time domain waveform diagram of the demodulated 4-channel pulse wave signal and the synthesized signal in Experiment 1 of this application. From top to bottom, the demodulated signals and synthesized waveforms of channel 1, channel 2, channel 3, and channel 4 are shown.
[0037] Figure 5 This is a comparison diagram of the amplitude and frequency spectrum of the 4-channel pulse wave signal, the synthetic signal and the ECG signal in Experiment 2 of this application. From top to bottom, it is the amplitude and frequency spectrum of the demodulated signal, the synthetic signal and the detected ECG of channel 1, channel 2, channel 3, channel 4.
[0038] Figure 6This is a comparison diagram of the amplitude and frequency spectrum of the 4-channel pulse wave signal, the synthetic signal and the ECG signal in Experiment 3 of this application. From top to bottom, it is the amplitude and frequency spectrum of the demodulated signal, the synthetic signal and the detected ECG of channel 1, channel 2, channel 3, channel 4.
[0039] Figure 7 The waveform diagram before demodulation of this application is shown in FIG. 1 , 2 , 3 , and 4 from top to bottom.
[0040] Figure 8 The following is a schematic diagram of the waveform after demodulation in this application. From top to bottom, they are the demodulated signal waveforms of channel 1, channel 2, channel 3, and channel 4. DETAILED DESCRIPTION
[0041] Embodiment. In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application, that is, the embodiments described are only part of the embodiments of the present application, not all of the embodiments.
[0042] The following is combined with Figures 1 to 8 Further detailed description of this application.
[0043] An intelligent wearable heart rate detection device, which includes LED, PD, IV conversion circuit, AD acquisition circuit, processor and program, such as Figure 1 This embodiment adopts an optimized multi-LED, multi-PD probe layout to obtain a larger AC / DC ratio for the PPG signal, and designs a corresponding IV conversion circuit and a signal acquisition scheme; the collected multi-channel photoelectric conversion signals are demodulated using a phase-locked amplification technology to obtain multi-channel pulse wave signals; for the demodulated multi-channel pulse wave signals, an optimization problem is defined and synthesis enhancement processing is performed on them.
[0044] The specific layout of the probe in this embodiment is as follows Figure 2As shown in the figure, LED1, LED2, LED3, and LED4 are four SFH 7018A multicolor LED package products, each SFH 7018A has three LEDs: green, red, and infrared; PD1, PD2, PD3, and PD4 are four SFH 2706 small-size photodetectors. According to the selected LED and PD models, in order to obtain a larger AC / DC ratio of the measured PPG signal, the geometric center of the four LED devices is set on a circle with a radius of 4.475 mm. Since the penetration depth of green light is shallow, the distance between the green light source and the PD is smaller than that of red light and infrared light, that is, the green LED is closer to the center of the circle, so that the AC / DC ratio of the green light PPG signal is maximized; at the same time, the geometric center of the effective photosensitive area of the four PDs is set on a concentric circle with a radius of 3.6 mm, and the LED devices and PDs are kept staggered. The central angle corresponding to the geometric center of each LED device and the geometric center of the effective photosensitive area of the adjacent PD is 45°.
[0045] The circuit design of this embodiment matches the probe solution, and the specific principle is as follows: after the circuit is powered on, the host computer sets the detection mode, light frequency, sampling rate, etc. and sends them to the single-chip microcomputer STM32F401. The STM32F401 outputs a PWM wave with a set frequency to simultaneously control the on and off of four NMOS tubes to drive the corresponding color LEDs in the four SFH 7018A to flash. The reflected light emitted by the LED after passing through the human tissue is detected by the four photodetectors SFH 2706, and the output photocurrent passes through the IV conversion circuit built with OPA2325, and then is connected to the AD7606 sampling ADC.
[0046] The data acquisition scheme of this embodiment matches the probe scheme. The ADC sampling rate is set to 10kHz, and the green light flashes at a frequency of 500Hz. AD7606 synchronously collects 7-channel signals. Channels 1 to 4 collect 4 voltage signals after IV conversion, and channels 5 to 7 respectively collect PWM waves for driving red, infrared, and green LEDs.
[0047] In this embodiment, the multi-channel photoelectric conversion signals are demodulated by using the phase-locked amplification technology. The signals of channels 1 to 4 collected by the ADC are multiplied by channel 7 corresponding to the green light flashing PWM wave. Then, the calculated 4-channel signals are respectively passed through a 2nd-order Butterworth low-pass filter with a cutoff frequency of 0.7 Hz and a 3rd-order Butterworth high-pass filter with a cutoff frequency of 5 Hz for signal preprocessing to filter out high-frequency noise and baseline drift. The waveforms of the 4-channel signals before demodulation are as follows: Figure 7 As shown, after demodulation, the 4-channel pulse wave signal waveform is as follows Figure 8 As shown, the signal quality is good and consistent with the PPG waveform characteristics.
[0048] In this embodiment, an optimization problem is defined for the multi-channel pulse wave signals obtained by demodulation, and synthesis enhancement processing is performed on them to find a set of synthesis coefficients a, b, c, and d with a square sum of 1. The above coefficients are respectively multiplied with the PPG signals of the four channels and then added to obtain the synthesized enhanced PPG signal; the objective function is defined as the sum of the amplitudes corresponding to the spectrum peak and its second harmonic and third harmonic in the amplitude spectrum obtained by FFT of the synthesized signal; in order to find a set of normalized synthesis coefficients that maximize the objective function, the four coefficients are traversed between -1 and 1 with an interval of 0.2 through a grid search method, and divided by their square sum to achieve normalization, so as to find a set of coefficients that maximize the objective function: this set of coefficients is used as the initial point of nonlinear optimization, and a set of coefficients that maximize the objective function is found in the range of -1 to 1, and this set of coefficients is used to synthesize the enhanced PPG signal; for the synthesized enhanced PPG signal, the frequency value corresponding to the spectrum peak in the spectrum is multiplied by 60 to obtain the average heart rate value during detection.
Claims
1. An intelligent wearable heart rate detection device, comprising LED, PD, IV conversion circuit, AD acquisition circuit, processor and program, characterized in that: (1) An optimized multi-LED and multi-PD probe layout is adopted to obtain a larger AC-DC ratio of the PPG signal, and a corresponding multi-channel IV conversion circuit and signal synchronization acquisition scheme are adopted; (2) The multi-channel photoelectric conversion signals are demodulated using the phase-locked amplification technology to obtain multi-channel pulse wave signals; (3) An optimization problem is defined for the demodulated multi-channel pulse wave signals, and synthesis enhancement processing is performed on them, and then the heart rate is calculated based on the enhanced signals.
2. According to claim 1, a smart wearable heart rate detection device adopts an optimized multi-LED, multi-PD probe layout, characterized in that: Four multicolor LED package products and four photodetectors are selected. Each multicolor LED package integrates three wavelengths of light: green, red and infrared. The geometric centers of the four LED package products and the geometric centers of the effective photosensing areas of the four PDs are located on two concentric circles with different radii. The LED devices and PDs are staggered, and the central angle between the geometric center of each LED device and the geometric center of the effective photosensing area of the adjacent PD is kept at 45°. In order to obtain a better maximum AC / DC ratio of the PPG signal, the green LED is kept closer to the center of the circle than the red and infrared LEDs, and the distance between the LED and the PD is appropriate.
3. According to claim 1, the circuit design of the intelligent wearable heart rate detection device should match the probe solution, characterized in that: The circuit includes LED driving circuit, IV conversion circuit, signal acquisition circuit, etc. The single-chip microcomputer generates a PWM wave of specified frequency to control the on and off of the NMOS tube, thereby controlling the flashing frequency of the LED; the heart rate mode can be selected on the host computer to make the four green LEDs flash at the same time at a specified frequency to measure the heart rate, or the blood oxygen mode can be selected to make the four red LEDs and the four infrared LEDs flash alternately at a specified frequency; the IV conversion circuit converts the photocurrent output by the photodetector into a voltage signal of appropriate magnitude, and the voltage signal is collected by the ADC acquisition circuit.
4. According to claim 1, the data collection of the intelligent wearable heart rate detection device should match the probe solution, characterized in that: The original data of 7 channels are collected synchronously. Channels 1 to 4 are the voltages after the photocurrents of PD1 to PD4 are converted through IV. Channels 5 to 7 correspond to the PWM waves driving red light, infrared light, and green light, respectively.
5. According to the intelligent wearable heart rate detection device of claim 1, the multi-channel photoelectric conversion signals are demodulated by using the phase-locked amplification technology, characterized in that: The 4-channel PPG raw data are multiplied by the PWM wave driving the green LED and then filtered to remove high-frequency noise and baseline drift to obtain a 4-channel pulse wave signal.
6. According to the intelligent wearable heart rate detection device of claim 1, an optimization problem is defined for the demodulated multi-channel pulse wave signals, and synthesis enhancement processing is performed on them, characterized in that: The objective function of the optimization problem is defined as the sum of the highest spectral peak of the synthetic signal spectrum and its second harmonic and third harmonic amplitudes, in order to find a set of normalized combination vectors that maximize the objective function and synthesize and enhance the four-channel pulse signals obtained by phase-locked amplification.