A PPG-based head-mounted heart rate detection method, system and head-mounted device

Abstract

The application provides a PPG-based head-mounted heart rate detection method, a heart rate detection system for implementing the method and a head-mounted device. The PPG-based head-mounted heart rate detection method comprises the following steps: collecting PPG and ACC data at a forehead vein; data preprocessing; adopting the adaptive motion mode heart rate detection method as claimed in claim 1; and heart rate post-processing. The motion mode related heart rate detection technology suitable for real-time detection is provided, different heart rate extraction optimization algorithms can be selected according to different motion modes, the implementation complexity of the algorithm is reduced, and the accuracy of heart rate detection is improved.

Description

Technical Field

[0001] This invention belongs to the field of heart rate detection technology, and particularly relates to a head-mounted heart rate detection method, system and head-mounted device based on PPG. Background Technology

[0002] PPG is a non-invasive detection method that uses photoelectric techniques to detect changes in blood volume in living tissue. Since ECG electrodes need to be placed on both sides of the heart and in close contact with the skin, prolonged wear can significantly reduce the user experience. PPG, on the other hand, can be integrated into devices such as wristbands, watches, and smart helmets, making it convenient for users and providing an excellent experience. Therefore, in recent years, there has been an increasing amount of research on extracting heart rate from PPG signals.

[0003] Existing heart rate detection methods are all based on processing wristband PPG data. However, wrist motion artifacts are larger than those on the forehead, requiring complex algorithms to remove motion artifacts. In addition, existing studies have not evaluated PPG heart rate detection under different sports scenarios.

[0004] In view of this, it is indeed necessary to design a head-mounted heart rate detection method, system, and device to solve the above problems. Summary of the Invention

[0005] The main objective of this invention is to propose a heart rate detection technology suitable for real-time detection under exercise modes. Different heart rate extraction optimization algorithms can be selected according to different exercise modes to reduce the implementation complexity of the algorithms and improve the accuracy of heart rate detection.

[0006] To achieve the above objectives, this invention provides an adaptive exercise mode heart rate detection method. Based on PPG and ACC data, an SVM model is used to classify exercise modes into categories such as stationary, walking, running, and climbing stairs. For stationary modes, FFT is used to calculate the spectrum of PPG data, and then the frequency corresponding to the maximum value in the spectral peak is multiplied by 60 as the baseline heart rate. For scenarios where the maximum spectral peak frequency of acceleration and the maximum spectral peak frequency of PPG are easily distinguishable, such as running mode, an optimized spectral subtraction method is used to remove motion artifacts. First, the spectra of PPG data and triaxial acceleration composite data are calculated using FFT. Then, the two frequency points with larger amplitude values ​​in the PPG spectrum are found, and the values ​​closest to the maximum peak frequency of acceleration are removed. The remaining frequency is used to calculate the heart rate, which is ppg_freq*60. Finally, the final heart rate value is output after heart rate correction. For scenarios where the maximum spectral peak frequency of acceleration and the maximum spectral peak frequency of PPG are not easily distinguishable, such as walking and climbing stairs, an optimized RLS adaptive filtering method is used to remove motion artifacts. First, the Pearson correlation coefficient ρ of PPG, ACC, and PPG, Gyro is calculated respectively. pa ρ pg Find the maximum value of both ρ = max(ρpa , p pg ), if p < e, the PPG signal adopts the optimized spectrum subtraction to remove the motion artifact, if p >= e, the PPG signal adopts the RLS adaptive filter to eliminate the motion artifact, wherein the reference signal of the adaptive filter uses the ACC or Gyro signal corresponding to p;then the heart rate value is calculated again through the frequency domain peak value tracking;finally, the final heart rate value is output through the heart rate correction.

[0007] To achieve the above purposes, the application further provides a head-mounted heart rate detection method based on PPG, comprising the following steps:

[0008] Step one, collecting PPG and ACC data at the forehead vein;

[0009] Step two, data preprocessing;

[0010] Step three, adopting an adaptive motion mode heart rate detection method;and

[0011] Step four, heart rate post-processing.

[0012] Further improvement of the application is that a non-invasive PPG sensor and an acceleration sensor are used to collect PPG and ACC data at the forehead vein, and a POLAR H10 ECG sensor is used to record ECG data at the chest of the subject as a reference for comparison.

[0013] Further improvement of the application is that in the step two, the data is segmented by using the sliding window method;since the normal heart rate value is between 40-220 bpm, the segmented PPG and ACC data are filtered through the Butterworth band-pass filter to filter out the noise other than the heart rate frequency.

[0014] Further improvement of the application is that in the step four, the heart rate value is corrected according to the continuity of the heart rate value change of the normal healthy population and the acceleration data to judge the motion intensity.

[0015] To achieve the above purposes, the application further provides a head-mounted adaptive heart rate detection method, which is used for a head-mounted device and comprises the following steps:

[0016] First step, judging whether the measured person wears the sensor, if yes, entering the next step, if not, entering the low-power consumption mode;

[0017] Second step, entering the working mode, the working mode is the heart rate detection method as described above.

[0018] To achieve the above object, the application further provides a head-mounted device for detecting heart rate, comprising a microprocessor, a PPG heart rate sensor, an acceleration sensor and a Bluetooth module, and the head-mounted device can implement the method of any one of the preceding.

[0019] To achieve the above object, the application further provides a PPG-based head-mounted heart rate detection system, comprising a data acquisition module, a data preprocessing module, a detection module capable of implementing the adaptive motion mode heart rate detection method of the preceding, and a heart rate correction module.

[0020] The application is further improved in that a non-invasive PPG sensor and an acceleration sensor are worn at the forehead vein to collect PPG and ACC data, and a POLAR H10 ECG sensor is worn at the chest of the subject to record ECG data as a reference for comparison.

[0021] The application is further improved in that the data preprocessing module adopts a sliding window method to segment the data; since the normal heart rate value is between 40-220 bpm, the segmented PPG and ACC data are filtered through a Butterworth band-pass filter to filter out noise other than the heart rate frequency.

[0022] The application has the following beneficial effects: the motion mode related heart rate detection technology suitable for real-time detection according to the application selects different heart rate extraction optimization algorithms according to different motion modes, thereby reducing the implementation complexity of the algorithm. Meanwhile, placing the heart rate detection in a helmet can not only save wristband devices, but also improve the heart rate detection accuracy in the corresponding motion mode. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 The figure is a hardware configuration diagram of the head-mounted device of the application.

[0024] Figure 2 The figure is a flowchart of the PPG-based head-mounted heart rate detection method of the application.

[0025] Figure 3 The figure is a flowchart of the adaptive motion mode heart rate detection of the application.

[0026] Figure 4 The figure is a schematic diagram of the relative position of the PPG sensor on the forehead artery and vein of the human body.

[0027] Figure 5 The figure is a schematic diagram of the position of the POLAR H10 heart rate belt on the human body. DETAILED DESCRIPTION

[0028] In order to make the object, technical scheme and advantages of the application clearer, the application is described in detail below with reference to the drawings and specific embodiments.

[0029] It should be emphasized that in the description of the present application, various formulas and constraints are distinguished by consistent labels before and after, but it is not excluded that different labels are used to mark the same formula and / or constraint, and the purpose of such setting is to make the features of the present application more clear.

[0030] The existing heart rate detection method is based on wristband PPG data processing, and the wrist movement relative to the forehead has large artifacts, which requires complex motion artifact removal algorithm. In addition, the existing research does not study and evaluate PPG heart rate detection under different motion scenarios. Therefore, placing heart rate detection in the intelligent helmet not only saves the wristband device, but also improves the accuracy of heart rate detection under the corresponding motion mode. The PPG-based head-mounted heart rate detection technology proposed in the present application includes a motion mode related heart rate detection technology suitable for real-time detection and an optimization algorithm based on the PPG head-mounted heart rate detection technology. The motion mode is divided into static, walking, running, going up and down stairs, etc. using the SVM model, and different optimization algorithms are used for different motion modes, so as to effectively calculate the real-time heart rate.

[0031] The present application will be specifically described below in combination with the drawings.

[0032] The hardware solution of the system of the present application includes a microprocessor MCU using a single-chip microcomputer as a main control chip, a data acquisition part, and a communication part. The data acquisition part is a PPG heart rate sensor and an acceleration sensor. The communication part can use a Bluetooth module. The hardware model is as shown in Figure 1 .

[0033] The low-power research of the head-mounted device mainly includes two parts. The first part is the automatic sleep and wake-up function of the single-chip microcomputer, and the second part is the self-adaptive motion mode heart rate extraction algorithm. The system software flow chart is as shown in Figure 2 .

[0034] The traditional heart rate extraction method does not study and evaluate PPG heart rate detection under different motion scenarios, and is not suitable for real-time processing of single-chip microcomputers. The self-adaptive motion mode extraction algorithm proposed in the present application solves the above problems by introducing the SVM model. As shown in Figure Three , the method mainly includes four modules / stages: first, the data acquisition stage; second, the pretreatment stage, which uses the sliding window method to segment the data; third, the head-mounted heart rate detection algorithm; fourth, the heart rate correction.

[0035] First, the data acquisition stage. Non-invasive PPG sensors and acceleration sensors are worn on the forehead vein to collect PPG and ACC data, and at the same time, POLAR H10 ECG sensors are worn on the chest of the subject to record ECG data as a reference for comparison.

[0036] The second step, the preprocessing stage, involves segmenting the data using a sliding window method. Since normal heart rate values ​​are between 40-220 bpm, the segmented PPG and ACC data are passed through a Butterworth bandpass filter to remove noise other than the heart rate frequency.

[0037] The third step involves the head-mounted heart rate detection algorithm. This paper uses an SVM model to classify exercise modes, mainly into stationary, walking, running, and stair climbing. For stationary modes, FFT is used to calculate the spectrum of PPG data, and then the frequency corresponding to the maximum value in the spectral peak is multiplied by 60 as the baseline heart rate. For scenarios where the maximum spectral peak frequency of acceleration and the maximum spectral peak frequency of PPG are easily distinguishable, such as running mode, an optimized spectral subtraction method is used to remove motion artifacts. First, FFT is used to calculate the spectrum of PPG data and triaxial acceleration composite data respectively; then, the two frequency points with larger amplitude values ​​in the PPG spectrum are found, and the values ​​closest to the maximum peak frequency of acceleration are removed. The remaining frequency is used to calculate the heart rate, which is ppg_freq*60; finally, the final heart rate value is output after heart rate correction. For scenarios where the maximum spectral peak frequency of acceleration and the maximum spectral peak frequency of PPG are not easily distinguishable, such as walking and stair climbing modes, an optimized RLS adaptive filtering method is used to remove motion artifacts; first, the Pearson correlation coefficient ρ of PPG, ACC and PPG, Gyro is calculated respectively. pa ρ pg Find the maximum value of both ρ = max(ρ pa , ρ pg If ρ < ε, the PPG signal uses optimized spectral subtraction to remove motion artifacts; if ρ ≥ ε, the PPG signal uses an RLS adaptive filter to eliminate motion artifacts. The reference signal for the adaptive filter uses the ACC or Gyro signal corresponding to ρ. Then, the heart rate value is calculated by frequency domain peak tracking. Finally, the final heart rate value is output through heart rate correction.

[0038] The fourth step is heart rate post-processing (correction). Since the heart rate changes in normal healthy individuals are continuous and acceleration data can be used to determine exercise intensity, these two characteristics can be used to correct the heart rate estimate.

[0039] like Figure 4 As shown, the PPG sensor is placed at the arterial vein to collect signals.

[0040] like Figure 5 As shown, the reference heart rate data was collected using a POLAR H10 at the chest position.

[0041] The application is based on wristband and head-mounted signal feature research, and proposes a head-mounted heart rate detection technology based on PPG. On the one hand, the SVM model is innovatively introduced to predict the motion mode, and a motion mode related heart rate detection technology suitable for real-time detection is proposed. On the other hand, for the characteristics of head-mounted motion artifacts, an optimization algorithm for the head-mounted heart rate detection technology based on PPG is innovatively proposed, mainly including the optimization of motion artifact spectrum subtraction in running state and the dynamic selection of RLS adaptive filter reference signal based on PPG signal and acceleration and angular velocity Pearson correlation coefficient in walking state to remove motion artifacts.

[0042] Experiments show that under the mode of daily activities, the mean absolute error (MAE) of head-mounted PPG heart rate detection is lower than that of wristband. The current PPG-based heart rate extraction algorithm does not study and evaluate PPG heart rate detection under different motion scenes, so the motion mode related heart rate detection technology suitable for real-time detection proposed in the application selects different heart rate extraction optimization algorithms according to different motion modes, reduces the implementation complexity of the algorithm, and improves the accuracy of heart rate detection.

[0043] In addition, the helmet is a kind of equipment for protecting the head, which is widely used in traffic driving, military, sports, work and other aspects. With the implementation and promotion of the national "one helmet and one belt" policy, the demand for helmets has increased, and the production of helmets has also increased. In 2021, the output of the helmet industry in China increased to 5163.2 million, and the demand for the helmet industry increased to 4640.9 million. Smart helmets can also be widely used in cycling competitions. The organizers can monitor the heart rate changes and brain signals of each cycling player at any time. In addition, in some industries such as coal mines, workers need to operate underground, and an intelligent helmet that can monitor heart rate at any time is undoubtedly very important. Based on this background, the head-mounted adaptive heart rate detection scheme proposed in the application has great research value. In the future, combined with electroencephalogram data, it will have a broad application space.

[0044] The above examples are only used to illustrate the technical solutions of the application and are not limiting. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the application.

Claims

1. An adaptive motion pattern heart rate detection method based on PPG data, ACC data, characterized in that: SVM model is used to classify the motion patterns, including static, walking, running, going up and down stairs, etc. For the static mode, FFT is used to calculate the spectrum of PPG data, and then the maximum value of the spectrum peak is found to correspond to the frequency * 60 as the reference heart rate. For the scenes such as running mode which are easy to distinguish the maximum spectrum peak frequency of acceleration and the maximum spectrum peak frequency of PPG, an optimized spectral subtraction is used to remove the motion artifact. First, FFT is used to calculate the spectrum of PPG data and three-axis acceleration combined data respectively. Then, two frequency points with larger amplitude values in the PPG spectrum are found, and the value closest to the maximum peak frequency point of acceleration is removed. The remaining frequency is used to calculate the heart rate, i.e. ppg_freq*60. Finally, the final heart rate value is output through heart rate correction. For the scenes such as walking and going up and down stairs which are not easy to distinguish the maximum spectrum peak frequency of acceleration and the maximum spectrum peak frequency of PPG, an optimized RLS adaptive filter is used to remove the motion artifact. First, the Pearson correlation coefficients ρ pa , ρ pg of PPG, ACC and PPG, Gyro are calculated respectively, and the maximum value ρ = max(ρ pa , ρ pg ) is found. If ρ < ε, the optimized spectral subtraction is used to remove the motion artifact of PPG signal, and if ρ ≥ ε, the RLS adaptive filter is used to remove the motion artifact of PPG signal, wherein the reference signal of the adaptive filter uses the ACC or Gyro signal corresponding to ρ. Then, the heart rate value is calculated through frequency domain peak value tracking. Finally, the final heart rate value is output through heart rate correction.

2. A PPG-based head-mounted heart rate detection method, characterized in that, The method comprises the following steps: Step 1: collecting PPG and ACC data at the forehead vein; Step 2: data preprocessing; Step 3: using the adaptive motion mode heart rate detection method as claimed in claim 1; and Step 4: heart rate post-processing.

3. The PPG-based head-mounted heart rate detection method of claim 2, wherein: The non-invasive PPG sensor and the acceleration sensor are worn at the forehead vein to collect PPG and ACC data, and the POLAR H10 ECG sensor is worn on the chest of the subject to record ECG data as a reference for comparison.

4. The PPG-based head-mounted heart rate detection method of claim 2, wherein: In the step 2, the data is segmented by using the sliding window method; since the normal heart rate value is between 40-220 bpm, the segmented PPG and ACC data are filtered by the Butterworth band-pass filter to filter out the noise other than the heart rate frequency.

5. The PPG-based head-mounted heart rate detection method of claim 2, wherein: In the step 4, the heart rate value is corrected according to the continuity of the heart rate value change of the normal healthy people and the acceleration data to judge the motion intensity.

6. A head-mounted adaptive heart rate detection method for a head-mounted device, characterized in that: Step 1: judging whether the subject wears the sensor, if yes, entering the next step, if not, entering the low-power consumption mode; Step 2: entering the working mode, the working mode being the heart rate detection method as claimed in claim 2.

7. A headgear usable in detecting heart rate, characterized in that: The head-mounted device comprises a microprocessor, a PPG heart rate sensor, an acceleration sensor, and a Bluetooth module, and can implement the heart rate detection method as claimed in any one of claims 1-6.

8. A PPG-based head-mounted heart rate detection system, characterized in that, The head-mounted device comprises a data acquisition module, a data preprocessing module, a detection module capable of executing the adaptive motion mode heart rate detection method as claimed in claim 1, and a heart rate correction module.

9. The PPG-based head-mounted heart rate detection system of claim 8, wherein: The non-invasive PPG sensor and the acceleration sensor are worn at the forehead vein to collect PPG and ACC data, and the POLAR H10 ECG sensor is worn on the chest of the subject to record ECG data as a reference for comparison.

10. The PPG-based head-mounted heart rate detection system of claim 9, wherein: The data preprocessing module segments the data by using the sliding window method; since the normal heart rate value is between 40-220 bpm, the segmented PPG and ACC data are filtered by the Butterworth band-pass filter to filter out the noise other than the heart rate frequency.

Images