Physiological parameter detection method adaptive to individual difference factors

Abstract

The invention relates to the technical field of physiological parameter detection, in particular to a physiological parameter detection method adaptive to individual difference factors, which comprises the following steps: (1) acquiring an electrocardiosignal and a PPG signal, and respectively performing primary filtering, primary high-frequency filtering and secondary filtering on the signals; (2) extracting an RR interval, a pulse rate, a heart rate and heart rate pulse conduction time based on the signal after the secondary filtering processing; (3) processing a differential signal corresponding to the electrocardiosignal through a first window filter coefficient, and calculating q RR interval mean values of the first window filter coefficient; (4) processing the PPG signal through a second window coefficient to obtain q differential values and pulse rates of the second window coefficient; (5) obtaining the heart rate; and (6) calculating the R wave position within the range of 20-50ms after each frame of electrocardio is finished by using the electrocardio signal, and calculating the corresponding pulse conduction time, so that the individual adaptability of physiological parameter detection is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of physiological parameter detection, and in particular to a physiological parameter detection method that is adaptive to individual difference factors. Background Art

[0002] In recent years, with the increasing attention paid to health management and the rapid development of wearable devices and mobile medical technologies, real-time monitoring of physiological parameters has become an important research direction in the field of medical health. As important indicators reflecting the state of the human cardiovascular system, the accurate detection and analysis of ECG signals and photoplethysmogram (PPG) signals are of great significance for the early diagnosis and prevention of cardiovascular diseases.

[0003] However, existing physiological parameter detection methods often have difficulty maintaining stable and accurate performance in the face of individual differences and complex environmental interference. The invention patent with patent number 202010180921.6 proposes a real-time heart rate estimation method for dynamic electrocardiogram based on MODWT. Although this method improves the accuracy of heart rate estimation to a certain extent, the following problems still exist:

[0004] 1. This method fails to fully utilize the rich information contained in the ECG signal, resulting in the accuracy still needing to be improved in some complex situations, such as arrhythmia.

[0005] 2. Although adaptive filtering technology is used, it fails to effectively optimize parameters based on the physiological characteristics of different individuals, resulting in insufficient adaptability when facing a large range of people.

[0006] 3. In the case of large interference such as motion, the anti-interference ability of this method is limited and it is easily affected by motion artifacts, which affects the reliability of the measurement results.

[0007] 4. This method mainly focuses on the estimation of heart rate, but fails to provide effective measurement and analysis methods for other important physiological parameters, such as pulse arrival time (PAT) and pulse propagation time (PPT).

[0008] In view of these problems existing in the prior art, there is an urgent need for a method that can effectively adapt to individual differences, has strong anti-interference ability, and can accurately measure multiple physiological parameters at the same time. The present invention is proposed to meet this demand. Summary of the invention

[0009] The present invention provides a physiological parameter detection method which is self-adaptive to individual difference factors, aiming to solve the technical problems existing in the prior art such as insufficient individual adaptability, limited anti-interference ability, and low multi-parameter measurement accuracy.

[0010] The present invention provides a physiological parameter detection method that is self-adaptive to individual difference factors, comprising the following steps:

[0011] (1) obtaining an ECG signal and a PPG signal, and performing primary filtering, primary high-frequency filtering, and secondary filtering on the signals respectively;

[0012] (2) extracting RR interval, pulse rate, heart rate and heart rate pulse conduction time based on the signal after the secondary filtering process;

[0013] The specific steps of the secondary filtering include:

[0014] a) Using the first-order difference method to obtain differential signals of ECG and PPG respectively;

[0015] b) performing low-pass and high-pass filtering on the differential signal through a plurality of windows, and then performing high-pass and low-pass filtering;

[0016] c) taking the median of the differential and the original signal, performing median filtering, and finally taking the median of the differential;

[0017] d) determining the window filter coefficient according to different physiological parameter characteristics of different individuals;

[0018] (3) processing the differential signal corresponding to the ECG signal by the first window filter coefficient, and calculating the mean of q RR intervals of the first window filter coefficient;

[0019] (4) processing the PPG signal through the second window coefficient to obtain q differential values ​​of the second window coefficient and the pulse rate;

[0020] (5) dividing the output mean of the q pulse rate filters of the second window coefficient by the q mean values ​​RR of the first window coefficient to obtain the heart rate;

[0021] (6) Use the ECG signal to calculate the R wave position within 20-50ms after the end of each ECG frame, and calculate the corresponding pulse conduction time.

[0022] Specifically, the first-level filtering is used for high-frequency and high-frequency noise processing of ECG signals and PPG signals;

[0023] The first-level high-frequency filtering is to set the R wave amplitude of the original ECG signal to zero;

[0024] The first-level low-frequency noise filtering includes high-pass filtering and Gaussian filtering, wherein the high-pass filtering is used to remove the influence of human myoelectric interference on ECG and PPG signals when the body surface is stationary, and the Gaussian filtering is used to remove the influence of human breathing on ECG and PPG.

[0025] Specifically, the secondary low-frequency filtering includes median filtering and average filtering, specifically, the signal after the first-stage filtering is subjected to median filtering and average filtering of multiple windows;

[0026] The second window low-pass and high-pass filtering is used to remove the influence of different individuals on the RR interval;

[0027] The second high-pass and low-pass filtering is a Gaussian band-pass filtering, which obtains the heart rate based on the PPG signal by extracting the high-frequency information of the PPG signal.

[0028] Specifically, the heart rate calculated after the secondary filtering is used to update the filter coefficient of the second window filter and to detect whether arrhythmia occurs;

[0029] The physiological parameter detection device includes: an STM32F4 system, a radio frequency part, a power supply part, and a memory operation platform.

[0030] Specifically, the method further comprises the following steps:

[0031] (1) obtaining a user individual ID, and determining filter and filter coefficient configuration parameters according to the user individual ID;

[0032] (2) inputting the first interwave interval mean and pulse rate obtained from the ECG signal and the PPG signal in the first window into a pre-trained linear prediction model to obtain an optimized coefficient vector of individual differences, which is used to update the filter coefficient of the second window;

[0033] The linear prediction model input is the mean value of the first inter-wave interval of the second window and the pulse rate, and the output is the optimization coefficient vector.

[0034] Specifically, the optimization coefficient vector and the filter coefficient of the first window jointly update the second window filter;

[0035] The heart rate is calculated in a second window, and the filter coefficient vector and the optimization coefficient vector are transmitted in different second windows.

[0036] Specifically, the heart rate is obtained by:

[0037] The R wave information is obtained through the first window, and the RR interval is calculated, and the differential information of the PPG wave is obtained. Low-pass and high-pass filtering is performed through the second window to extract the high-frequency information of the PPG. Then, the differential of the PPG pulse rate is obtained through the second window, and the mean of the low-pass signal and the pulse rate are calculated respectively, and finally the heart rate and average PAT are obtained.

[0038] Specifically, the input of the PAT is the R wave apex of the electrocardiogram in the interval 20 to 50 ms before the start of the PAT calculation time, and the output is PAT;

[0039] The filtering input PPG signal and ECG signal of the PPT are specifically as follows: PPT=TPT2-TPTE2, wherein TPT is the pulse cycle time, and TPTE is the pulse cycle time of the corresponding point of the ECG R wave.

[0040] Specifically, the following individual information collection steps are further included:

[0041] (1) filtering the ECG signal using a filter of the first window;

[0042] (2) calculating the heart rate of the filtered signal by using a segmented calculation method of the ECG R wave differential, and using a multi-window moving average method to obtain the heart rate calculated in each second window;

[0043] (3) calculating a slope value for the heart rate in each window and the corresponding second-order difference, and dynamically gaining the slope value using a gain value;

[0044] (4) After extracting the dynamic gain, the result is used to extract the extreme point, where the extreme point corresponds to the maximum gain.

[0045] Specifically, the dynamic gain process is as follows: the difference between the previous point position and the current point position of the center point of each second window is calculated, and multiplied by the gain factor k;

[0046] The slopes of the two adjacent extreme value points before and after the gain are defined as the critical slope, the last of the critical slopes is the maximum point and the minimum point, the gain values ​​of the maximum point and the last minimum point in the critical slope are determined as the starting and ending intervals respectively, and the average between the two end points of the interval is taken as the final result;

[0047] The extreme point is determined by comparing the difference value, and the extreme value is determined when the difference value is greater than the set threshold. For the extreme point whose difference value is greater than the set threshold and the point is also the extreme point of the slope value of the current point;

[0048] The information collection also includes a step of identifying noise interference of the ECG waveform. When two maximum value points appear in a window, the window behind this window is included in the range of noise judgment.

[0049] The present invention proposes an innovative psychological state assessment method, which includes a psychological state acquisition module, a psychological state assessment module, a data optimization module and a smart bracelet terminal. The smart bracelet collects multi-dimensional physiological and environmental data, combines deep learning technology for analysis, and realizes real-time and continuous assessment of the psychological state.

[0050] The present invention realizes high-quality processing and analysis of ECG and PPG signals through innovative multi-stage filtering, adaptive window technology and individual parameter optimization, thereby achieving the following beneficial effects:

[0051] 1. Significantly improve the individual adaptability of physiological parameter detection. By introducing adaptive windows and individualized parameter optimization mechanisms, the present invention can dynamically adjust detection parameters according to the physiological characteristics of different individuals, so that the detection results can maintain high consistency and accuracy among people of different ages, genders and health conditions.

[0052] 2. The system's anti-interference ability has been greatly enhanced. The innovative multi-stage filtering processing technology effectively eliminates various environmental noises and physiological interferences, so that the system can still provide stable and reliable measurement results even under complex conditions such as exercise.

[0053] 3. Realizes high-precision measurement of multiple physiological parameters at the same time. The present invention can not only accurately measure heart rate, but also precisely locate R waves, calculate important physiological indicators such as PAT and PPT, and provide rich data support for comprehensive assessment of cardiovascular health status.

[0054] 4. Improved real-time performance and reliability of the system. By optimizing the algorithm design and signal processing flow, the present invention achieves low-latency real-time processing while ensuring high accuracy, meeting the needs of continuous monitoring.

[0055] 5. Enhanced applicability and scalability of the method. The method of the present invention is flexible in design and is not only suitable for resting state measurements, but also maintains good performance in a variety of scenarios such as daily activities and light exercise, providing the possibility for long-term and continuous health monitoring.

[0056] In general, the method of the present invention not only solves the problems of the prior art, but also opens up a new way for personalized and precise monitoring of physiological parameters. It is expected to play an important role in clinical diagnosis, telemedicine, health management and other fields, and make positive contributions to improving medical quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a signal acquisition and preprocessing block diagram of the present invention;

[0058] Figure 2 It is a parameter extraction block diagram of the present invention;

[0059] Figure 3 is a block diagram of the adaptive optimization of the present invention;

[0060] Figure 4 It is a block diagram of accurate R wave positioning of the present invention;

[0061] Figure 5 It is a block diagram of arrhythmia detection of the present invention. DETAILED DESCRIPTION

[0062] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0063] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0064] The present invention provides a physiological parameter detection method that is self-adaptive to individual difference factors, which can effectively adapt to the physiological parameter characteristics of different individuals, thereby improving the accuracy and reliability of detection. The present invention will be described in detail below in conjunction with specific embodiments.

[0065] Please refer to Figure 1-5 First, the core of the present invention is to achieve accurate processing of ECG signals and photoplethysmography (PPG) signals through multi-stage filtering and adaptive window technology. Specifically, the method includes the following key steps:

[0066] Step S1: Signal acquisition and preliminary processing.

[0067] In this step, the ECG signal and the PPG signal are first obtained. Then, the two signals are subjected to primary filtering, primary high-frequency filtering and secondary filtering respectively. Among them, the primary filtering module 1 is mainly used to process the high-frequency and low-frequency noise of the ECG signal and the PPG signal. Preferably, in one embodiment of the present invention, the function of the primary high-frequency filtering module 2 is to set the amplitude of the R wave of the original ECG signal to zero, so as to effectively eliminate the influence of the R wave on subsequent processing.

[0068] Next, the first-level low-frequency noise filter module 3 includes two submodules: high-pass filtering and Gaussian filtering. Among them, the high-pass filtering submodule 3a is used to remove the influence of human electromyographic interference on ECG and PPG signals when the body surface is stationary. Usually, we choose a high-pass filter with a cutoff frequency of 0.5Hz, because the frequency of most electromyographic interference is lower than this value. The Gaussian filtering submodule 3b is mainly used to remove the influence of human breathing on ECG and PPG. In view of the fact that the normal breathing frequency of a normal person is about 0.2-0.3Hz, we usually choose a Gaussian bandpass filter with a center frequency of 0.25Hz and a bandwidth of 0.1Hz.

[0069] Step S2: Secondary filtering process.

[0070] Secondary filtering is one of the core steps of the present invention. First, the ECG and PPG signals are respectively subjected to first-order differential method to obtain differential signals. The first-order differential is used here instead of the higher-order differential because the first-order differential can effectively highlight the change trend of the signal without introducing too much high-frequency noise.

[0071] Then, these differential signals are subjected to low-pass and high-pass filtering through multiple windows, and then to high-pass and low-pass filtering. Preferably, in one embodiment of the present invention, the window size we use is 256 sampling points, which is an empirical value based on the periodic characteristics of most physiological signals. The cutoff frequencies of low-pass and high-pass filtering are set to 25Hz and 0.5Hz respectively, while the cutoff frequencies of high-pass and low-pass filtering are set to 1Hz and 20Hz respectively. These parameters are selected to suppress noise to the greatest extent while retaining useful information.

[0072] Next, take the median of the differential and the original signal and perform median filtering, and finally take the median of the differential. The purpose of this step is to further eliminate possible impulse noise and improve signal stability. The window size of the median filter is usually selected as 5-7 sampling points, which is a trade-off between smoothing effect and computational efficiency.

[0073] Finally, and most importantly, this method determines the window filter coefficient according to the different physiological parameter characteristics of different individuals. This is the core of achieving self-adaptation. By dynamically adjusting the filter coefficient, we can better adapt to the physiological characteristics of different individuals.

[0074] Step S3: Parameter extraction.

[0075] Based on the signal after secondary filtering, we can extract several important physiological parameters, including RR interval, pulse rate, heart rate and pulse conduction time (PAT).

[0076] First, the differential signal corresponding to the ECG signal is processed by the first window filter coefficient, and the mean of q RR intervals of the first window filter coefficient is calculated. Here, q is usually selected to be 10-20, because a too small q value may not effectively smooth short-term fluctuations, while a too large q value may mask important physiological changes.

[0077] Then, the PPG signal is processed by the second window coefficient to obtain q differential values ​​of the second window coefficient and the pulse rate. The size of the second window is usually selected to be 512 sampling points, which is determined based on the characteristics of the PPG signal and the information we need to extract.

[0078] Next, the output mean of the q pulse rate filters of the second window coefficient is divided by the q mean values ​​RR of the first window coefficient to obtain the heart rate. This method can effectively eliminate the influence of short-term fluctuations and provide a more stable and reliable heart rate estimation.

[0079] Finally, the ECG signal is used to calculate the R wave position within 20-50ms after each ECG frame ends, and the corresponding pulse conduction time is calculated. The 20-50ms range is chosen because the R wave usually appears within this time window, and this range is small enough to ensure the accuracy of positioning.

[0080] Step S4: Adaptive optimization.

[0081] In order to further improve the accuracy of detection, the present invention also introduces an adaptive optimization mechanism. First, the user's individual ID is obtained, and the filter and filter coefficient configuration parameters are determined by the ID. In this way, personalized parameter settings can be saved and loaded for different users.

[0082] Then, the first interwave interval mean and pulse rate obtained from the ECG signal and PPG signal in the first window are input into the pre-trained linear prediction model. The input of this model is the first interwave interval mean and pulse rate of the second window, and the output is the optimized coefficient vector. Through this model, we can obtain the optimized coefficient vector of individual differences, which is used to update the filter coefficient of the second window.

[0083] Preferably, in one embodiment of the present invention, the linear prediction model can be expressed as:

[0084] Y=AX+B

[0085] Among them, Y is the optimization coefficient vector, X is the input feature (including the mean of the first interwave interval and the pulse rate), and A and B are model parameters. This model is trained with a large amount of historical data and can capture the difference characteristics between different individuals.

[0086] It is worth noting that the optimization coefficient vector and the filter coefficient of the first window jointly update the filter of the second window. This dual update mechanism can achieve rapid adaptation to individual differences while maintaining system stability.

[0087] Step S5: Heart rate calculation.

[0088] The present invention adopts a unique heart rate calculation method. First, the R wave information is obtained through the first window, and the RR interval is calculated. Then, the differential information of the PPG wave is obtained, and low-pass and high-pass filtering is performed through the second window to extract the high-frequency information of the PPG. Next, the differential of the PPG pulse rate is obtained through the second window, and the low-pass signal and the mean of the pulse rate are calculated respectively. Finally, the heart rate and average PAT are obtained through this information.

[0089] The advantage of this method is that it makes comprehensive use of the information of ECG and PPG signals, which can provide a more stable and accurate heart rate estimation. Especially in the case of exercise or other interference, this method shows strong robustness.

[0090] Step S6: PAT and PPT calculation.

[0091] In the present invention, the calculation of PAT (pulse arrival time) also adopts an innovative method. The input of PAT is the R wave apex of the electrocardiogram in the interval 20 to 50 ms before the start of the PAT calculation time, and the output is the PAT value. The selection of this time window is based on a large amount of experimental data and can effectively capture the characteristics of the R wave.

[0092] For the calculation of PPT (pulse transit time), we used the following formula:

[0093] PPT=TPT2-TPTE2

[0094] Among them, TPT is the pulse cycle time, and TPTE is the pulse cycle time of the corresponding point of the ECG R wave. This calculation method takes into account the time difference between the ECG and PPG signals, and can more accurately reflect the propagation time of blood from the heart to the periphery.

[0095] Step S7: Collecting individual information.

[0096] In order to further improve the adaptive ability of the system, the present invention also includes detailed individual information collection steps. First, the ECG signal is filtered using the filter of the first window. Then, the heart rate is calculated by using the method of segmented calculation of the ECG R wave differential, and the multi-window moving average method is used to obtain the heart rate calculated in each second window.

[0097] Next, the slope values ​​are calculated for the heart rate and the corresponding second-order difference in each window, and the gain values ​​are used to dynamically gain these slope values. The dynamic gain process here is as follows: the difference between the previous point position and the current point position of the heart rate in each second window is calculated and multiplied by the gain factor k. Usually, the value range of k is 1.5-2.5, which is an empirical value based on a large amount of experimental data.

[0098] Finally, the result after extracting the dynamic gain is used to extract the extreme point, which corresponds to the maximum gain. In this process, we define the slope of the two adjacent extreme points before and after as the critical slope, where the last one is the maximum point and the minimum point. We use the gain values ​​of the maximum point and the last minimum point in the critical slope as the start and end intervals, respectively, and take the average between the two end points of the interval as the final result.

[0099] It is worth noting that the determination of extreme points is achieved by comparing the difference values. Specifically, when the difference value is greater than the set threshold (usually set to 1.5 times the average difference value), we believe that an extreme point has been found. At the same time, the point also needs to meet the condition that it is the extreme point of the slope value of the current point.

[0100] In addition, the present invention also includes a step of identifying noise interference of the ECG waveform. When two maximum points appear in a window, the window behind this window will be included in the scope of noise judgment. This method can effectively identify and process possible noise interference and improve the stability and reliability of the system.

[0101] In short, the physiological parameter detection method for adaptive individual difference factors provided by the present invention realizes accurate detection of physiological parameters of different individuals through innovative means such as multi-stage filtering processing, adaptive window technology, and individualized parameter optimization. This method can not only adapt to individual differences, but also has strong anti-interference ability, and has shown excellent performance in practical applications. Especially in scenarios where long-term and continuous monitoring of physiological parameters is required, the advantages of this method are more obvious. Through this method, we can provide more reliable and accurate data support for medical diagnosis, health monitoring and other fields, thereby laying a solid foundation for related research and applications.

[0102] Step S8: filter coefficient update and heart rate calculation.

[0103] In the present invention, the heart rate calculated after the secondary filtering is not only used for monitoring physiological parameters, but also used to update the filter coefficient of the second window filter. This dynamic update mechanism enables the system to continuously adapt to changes in the individual's physiological state. In addition, the heart rate value is also used to detect whether arrhythmia occurs. Preferably, in one embodiment of the present invention, when the change of three consecutive heartbeat cycles exceeds 20%, the system will determine that arrhythmia may occur and give a corresponding warning.

[0104] It is worth noting that the physiological parameter detection device of the present invention includes several key components: STM32F4 system, radio frequency part, power supply part and memory operation platform. Among them, the STM32F4 system is responsible for executing complex signal processing algorithms as the core processing unit. The radio frequency part is used for wireless transmission of data, so that the device can transmit the test results to the monitoring terminal in real time. The power supply part adopts a constant voltage linear power supply to ensure the stable operation of the system. The memory operation platform is used to store individualized parameter settings and historical data, which makes long-term monitoring possible.

[0105] Step S9: Accurate positioning of the R wave position.

[0106] The accurate positioning of the R wave position is crucial for the calculation of multiple physiological parameters. To this end, the present invention adopts a multiple determination method to ensure the accuracy of the R wave position.

[0107] First, the ECG signal after two-stage filtering is differentiated using the first window. Here, the width of the first window is selected in the range of 0.6-2 heartbeat intervals. This range is selected based on a large amount of experimental data and can effectively capture the characteristics of the R wave without introducing too much interference.

[0108] Then, the point where the differential change rate is less than the set threshold is determined as the suspected R wave position. Preferably, in one embodiment of the present invention, this threshold is set to 1.5 times the average differential change rate. The selection of this threshold can effectively identify the R wave and exclude most of the false positive results.

[0109] Next, the multi-window moving average method is used to solve the differential average, and the eigenvalues ​​are selected on the multi-window average differential curve. The eigenvalues ​​here include the position of the value and the two characteristics of width and narrowness. Usually, we choose a window size of 128 sampling points, which is a trade-off between computational efficiency and accuracy.

[0110] Based on the above characteristic values, we further determine the suspected R wave position. In order to further improve the accuracy, the present invention also introduces morphological feature determination. Specifically, if the change trend of the two points before and after the determined suspected R wave position is consistent with the suspected R wave, and the angle size is within a given threshold (usually set to 60°-120°), then the position is determined to be an R wave.

[0111] Finally, in order to deal with possible false positive or false negative results, the present invention also adopts a dual threshold detection method. When there are two suspected R wave positions that meet the above requirements, we finally determine it as the R wave position. This conservative strategy can significantly reduce the misjudgment rate and improve the reliability of the system.

[0112] Step S10: Accurate calculation of PAT and PPT

[0113] After determining the R wave position, we can perform accurate calculations of PAT (Pulse Arrival Time) and PPT (Pulse Transit Time). First, based on the determined R wave position, we obtain the pulse cycle time of the PPG signal.

[0114] Then, the R wave position is subtracted by 10-50ms to obtain the PAT corresponding to the R wave. The selection of this time range is based on physiological knowledge and a large amount of experimental data, and can effectively capture the changes in PAT.

[0115] Next, the position corresponding to the pulse peak of the PPG pulse signal is subtracted from the R wave position to obtain the PPT. It is worth noting that we consider the first 10-50ms of the R wave, the interval of 0.6-2 heartbeat intervals, or the interval of 4-6 heartbeat intervals as the possible range of PAT of the R wave. This flexible range definition enables the system to adapt to the changes in PAT in different individuals and different physiological states.

[0116] Preferably, in one embodiment of the present invention, we use the following formula to calculate PPT:

[0117] PPT=TPT2-TPTE2

[0118] Among them, TPT2 is the current pulse cycle time, and TPTE2 is the pulse cycle time corresponding to the current ECG R wave. This calculation method takes into account the time difference between the ECG and PPG signals, and can more accurately reflect the propagation time of blood from the heart to the periphery.

[0119] The physiological parameter detection method for adaptive individual difference factors provided by the present invention realizes accurate detection of physiological parameters of different individuals through innovative means such as multi-stage filtering, adaptive window technology, and individualized parameter optimization. This method can not only adapt to individual differences, but also has strong anti-interference ability, and has shown excellent performance in practical applications.

[0120] Especially in scenarios where long-term and continuous monitoring of physiological parameters is required, the advantages of this method are more obvious. Through this method, we can provide more reliable and accurate data support for medical diagnosis, health monitoring and other fields, thus laying a solid foundation for related research and applications.

[0121] In order to verify the superiority of the present invention, we designed a series of experiments, including one embodiment and three comparative examples, which are intended to comprehensively evaluate the performance of the present invention under different conditions, especially in terms of adapting to individual differences and anti-interference capabilities.

[0122] Embodiment 1: The method of the present invention

[0123] In this embodiment, we adopted a complete physiological parameter detection method that is adaptive to individual differences. This includes core technologies such as multi-level filtering, adaptive window technology, and individualized parameter optimization. We selected 50 volunteers of different ages, genders, and health conditions and conducted tests in both resting and light exercise states.

[0124] Comparative Example 1: Traditional fixed parameter method

[0125] In this comparative example, we used the traditional fixed parameter filtering method without considering individual differences and adaptive optimization. The filtering parameters were set according to the average level of the general population.

[0126] Comparative Example 2: Simple Adaptive Method

[0127] This comparative example adopts a simple adaptive method, which only adjusts the filter parameters according to the short-term changes of the signal, but does not take into account the individual's long-term characteristics and complex changes in physiological state.

[0128] Comparative Example 3: Machine Learning Method

[0129] In this comparative example, we used a machine learning-based method that trained the model with a large amount of historical data but did not have a real-time adaptive adjustment mechanism.

[0130] Test indicators and methods:

[0131] 1. Heart rate accuracy: Compare with professional electrocardiogram equipment and calculate the mean absolute error (MAE).

[0132] 2. R-wave positioning accuracy: Compare with the R-wave position marked by professional equipment and calculate the positioning accuracy.

[0133] 3. PAT measurement stability: Calculate the coefficient of variation (CV) of the PAT measurement values.

[0134] 4. Anti-motion artifact capability: Calculate the effective data ratio under mild motion conditions.

[0135] 5. Individual adaptability: The consistency of measurement results between different individuals is expressed by the intraclass correlation coefficient (ICC).

[0136] Test results:

[0137]

[0138] Analysis and discussion:

[0139] It can be clearly seen from the test results that the method of the present invention (Example 1) performs well in all test indicators and is significantly better than other comparative methods.

[0140] First, in terms of heart rate accuracy, the mean absolute error of the present invention is only 1.2 bpm, which is much lower than other methods. This is due to our multi-stage filtering and adaptive optimization technology, which can effectively eliminate various interferences and extract high-quality heart rate information. Especially in the case of large individual differences, the advantages of the present invention are more obvious.

[0141] The accuracy of R-wave positioning is as high as 99.5%, which fully proves the effectiveness of our R-wave detection algorithm. Accurate R-wave positioning is the basis for calculating multiple physiological parameters and is crucial for the subsequent calculation of PAT, PPT and other parameters.

[0142] In terms of PAT stability, the coefficient of variation of the present invention is only 2.3%, which is much lower than other methods. This shows that our method can provide stable and reliable PAT measurement results, which is of great significance for evaluating vascular elasticity and cardiovascular health.

[0143] The ability to resist motion artifacts is an important indicator to measure the reliability of the method in practical applications. The present invention can still maintain a 92.5% effective data ratio under mild motion, which means that our method can provide continuous and reliable monitoring even in daily activities.

[0144] Finally, the individual adaptability index (ICC = 0.95) fully demonstrated the excellent performance of the invention in adapting to different individual characteristics. This means that our method can provide consistent and reliable measurement results for both young and old people, healthy people and people with cardiovascular diseases.

[0145] It is worth noting that although the method based on machine learning (Comparative Example 3) performs well in some indicators, it lacks real-time adaptive capabilities and is difficult to cope with dynamic changes in individual states. In contrast, the method of the present invention has both strong adaptive capabilities and low computational complexity, and is more suitable for real-time, long-term physiological parameter monitoring.

[0146] In summary, this set of experimental results strongly proves the innovation and superiority of the present invention in the field of physiological parameter detection. It not only surpasses the existing methods in all technical indicators, but more importantly, it provides the possibility for personalized and precise health monitoring. This detection method that can adapt to individual differences is expected to play an important role in clinical diagnosis, telemedicine, health management and other fields, and open up new ways to improve medical quality and efficiency.

[0147] In addition, the method of the present invention has good scalability and adaptability. By adjusting the filtering parameters, optimizing the algorithm or adding new physiological parameters, we can easily apply this method to a wider range of physiological parameter detection fields, providing strong technical support for personalized medicine and precise health management.

[0148] It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A physiological parameter detection method that is adaptive to individual difference factors, characterized in that The following steps are involved: (1) obtaining an ECG signal and a PPG signal, and performing primary filtering, primary high-frequency filtering, and secondary filtering on the signals respectively; (2) extracting RR interval, pulse rate, heart rate and heart rate pulse conduction time based on the signal after the secondary filtering process; The specific steps of the secondary filtering include: a) Using the first-order difference method to obtain differential signals of ECG and PPG respectively; b) performing low-pass and high-pass filtering on the differential signal through a plurality of windows, and then performing high-pass and low-pass filtering; c) taking the median of the differential and the original signal, performing median filtering, and finally taking the median of the differential; d) determining the window filter coefficient according to different physiological parameter characteristics of different individuals; (3) processing the differential signal corresponding to the ECG signal by the first window filter coefficient, and calculating the mean of q RR intervals of the first window filter coefficient; (4) processing the PPG signal through the second window coefficient to obtain q differential values ​​of the second window coefficient and the pulse rate; (5) dividing the output mean of the q pulse rate filters of the second window coefficient by the q mean values ​​RR of the first window coefficient to obtain the heart rate; (6) Use the ECG signal to calculate the R wave position within 20-50ms after the end of each ECG frame, and calculate the corresponding pulse conduction time.

2. The method according to claim 1, characterized in that: The first-level filtering is used for high-frequency and high-frequency noise processing of ECG signals and PPG signals; The first-level high-frequency filtering is to set the R wave amplitude of the original ECG signal to zero; The first-level low-frequency noise filtering includes high-pass filtering and Gaussian filtering, wherein the high-pass filtering is used to remove the influence of human myoelectric interference on ECG and PPG signals when the body surface is stationary, and the Gaussian filtering is used to remove the influence of human breathing on ECG and PPG.

3. The method according to claim 1, characterized in that: The secondary low-frequency filtering includes median filtering and average filtering, specifically, the signal after the first-stage filtering is subjected to median filtering and average filtering of multiple windows; The second window low-pass and high-pass filtering is used to remove the influence of different individuals on the RR interval; The second high-pass and low-pass filtering is a Gaussian band-pass filtering, which obtains the heart rate based on the PPG signal by extracting the high-frequency information of the PPG signal.

4. The method according to claim 1, characterized in that: The heart rate calculated after the secondary filtering is used to update the filter coefficient of the second window filter and to detect whether arrhythmia occurs; The physiological parameter detection device includes: an STM32F4 system, a radio frequency part, a power supply part, and a memory operation platform.

5. The method according to claim 1, characterized in that: Further comprising the steps of: (1) obtaining a user individual ID, and determining filter and filter coefficient configuration parameters according to the user individual ID; (2) inputting the first interwave interval mean and pulse rate obtained from the ECG signal and the PPG signal in the first window into a pre-trained linear prediction model to obtain an optimized coefficient vector of individual differences, which is used to update the filter coefficient of the second window; The linear prediction model input is the mean value of the first inter-wave interval of the second window and the pulse rate, and the output is the optimization coefficient vector.

6. The method according to claim 5, characterized in that: The optimized coefficient vector and the filter coefficient of the first window jointly update the second window filter; The heart rate is calculated in a second window, and the filter coefficient vector and the optimization coefficient vector are transmitted in different second windows.

7. The method according to claim 1, characterized in that The specific method for obtaining the heart rate is: The R wave information is obtained through the first window, and the RR interval is calculated, and the differential information of the PPG wave is obtained. Low-pass and high-pass filtering is performed through the second window to extract the high-frequency information of the PPG. Then, the differential of the PPG pulse rate is obtained through the second window, and the mean of the low-pass signal and the pulse rate are calculated respectively, and finally the heart rate and average PAT are obtained.

8. The method according to claim 1, characterized in that: The input of the PAT is the R wave apex of the electrocardiogram in the interval 20 to 50 ms before the start of the PAT calculation time, and the output is PAT; The filtering input PPG signal and ECG signal of the PPT are specifically as follows: PPT=TPT2-TPTE2, wherein TPT is the pulse cycle time, and TPTE is the pulse cycle time of the corresponding point of the ECG R wave.

9. The method according to claim 1, characterized in that: The method further includes the following individual information collection steps: (1) filtering the ECG signal using a filter of the first window; (2) calculating the heart rate of the filtered signal using a segmented calculation of the ECG R wave differential method, and using a multi-window moving average method to obtain the heart rate calculated in each second window; (3) calculating the slope value for the heart rate in each window and the corresponding second-order difference, and dynamically gaining the above slope value using a gain value; (4) extracting the extreme point from the result after extracting the dynamic gain, and the extreme point corresponds to the maximum gain.

10. The method according to claim 9, characterized in that: The dynamic gain process is as follows: the difference between the previous point position and the current point position of the electrocardiogram in each second window is calculated, and the difference is multiplied by the gain factor k; the slope of the two adjacent extreme value points before and after the above gain is defined as the critical slope, the last of the critical slopes is the maximum point and the minimum point, the gain values ​​of the maximum point and the last minimum point in the critical slope are determined as the starting and ending intervals respectively, and the average between the two end points of the interval is taken as the final result; the extreme value point is determined by comparing the differential value, and the extreme value is determined when the differential value is greater than the set threshold, and the extreme value point of the slope value of the current point is also determined when the differential value is greater than the set threshold; the information collection also includes the step of identifying noise interference with the ECG waveform. When two maximum value points appear in a window, the window behind this window is included in the range of noise judgment.

Images