Human body electrocardio R wave detection method

Abstract

The invention discloses a human body electrocardio R wave detection method, and relates to the technical field of electrocardio signal detection.The system comprises the following components of gene expression data and electrocardio signal collection, specifically, gene expression level information of an individual whole genome is obtained through a gene detection technology, meanwhile, electrocardio signals are collected through a multi-channel electrocardio sensor, and the gene expression data and the electrocardio signals are obtained; according to the method, by introducing the gene expression data, personalized electrocardiogram R wave detection can be carried out on an individual with the tendency of specific gene variation or family genetic cardiovascular diseases, the conventional characteristics of the electrocardiogram signals are considered, information of the gene level is combined, and the method has the advantages of being high in accuracy and high in accuracy. Therefore, R waves can be recognized more accurately, false detection and missing detection are reduced, and the method has important significance in improving the accuracy of cardiovascular disease diagnosis, finding cardiovascular abnormality in an early stage and formulating a personalized treatment scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of electrocardiogram signal detection, and in particular to a method for detecting human electrocardiogram R waves. Background Art

[0002] ECG R-wave detection plays a vital role in the diagnosis of cardiovascular diseases, arrhythmia analysis and other fields. As a direct indicator of the electrophysiological activity of the heart, the accurate detection of ECG signals is of great significance for evaluating heart function and diagnosing heart disease. With the continuous advancement of medical technology, ECG R-wave detection methods are also constantly improving and perfecting, aiming to improve the accuracy and reliability of detection, thereby providing more accurate diagnosis and treatment options for patients with cardiovascular diseases.

[0003] Although the traditional ECG R wave detection method has achieved certain results, there are still some obvious limitations in practical applications. The traditional method is usually based on the recognition of the conventional characteristics of the ECG signal itself, and uses a general threshold or algorithm to detect the R wave. However, due to genetic differences, different individuals, especially those with specific gene mutations or a family tendency to cardiovascular disease, have significantly different cardiovascular physiological characteristics, which leads to large differences in the characteristics of the R wave in the ECG signal. The traditional general detection method is difficult to fully take into account these individual specificities, which often leads to inaccurate test results, which in turn affects the subsequent accurate diagnosis and assessment of cardiovascular diseases. In addition, the ECG signal is extremely susceptible to various noise interferences during the acquisition process, such as electromyographic noise, power frequency interference, etc. These noises will seriously interfere with the accuracy of R wave detection. The traditional detection algorithm lacks adaptive error correction capabilities. Once false detection or missed detection occurs, it can only output wrong results, and cannot be adjusted in real time according to the actual situation during the detection process to obtain the correct results. This is particularly unfavorable in the complex and changeable human physiological state and actual detection environment.

[0004] In view of the above problems, it is necessary to test the existing human ECG R-wave detection method, and realize personalized adjustment of the R-wave detection algorithm by in-depth analysis of the correlation between gene expression data and ECG signals. Therefore, it is of great significance to develop a human ECG R-wave detection method that can comprehensively realize the above characteristics. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the prior art and to provide a method for detecting R-waves in human ECG. The method can realize personalized adjustment of the R-wave detection algorithm by deeply analyzing the correlation between gene expression data and ECG signals, thereby significantly improving the accuracy and specificity of detection. At the same time, by adopting an advanced adaptive noise suppression and feature enhancement combined preprocessing method, the noise in the ECG signal is effectively suppressed and the R-wave features are enhanced, further improving the accuracy of detection. In addition, an adaptive fault-tolerant R-wave detection algorithm is developed to automatically identify and correct possible false detections, missed detections, etc. during the detection process, thereby ensuring that the detection algorithm maintains high accuracy and reliability in complex environments and changeable human physiological conditions.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for detecting human electrocardiogram R waves, the method comprising the following specific steps:

[0007] Gene expression data and ECG signal collection: Gene expression level information of the whole genome of an individual is obtained through gene detection technology, and ECG signals are collected using multi-channel ECG sensors. The time interval between the two collections is kept short, and the collected gene expression data and ECG signals are transmitted to the computer system;

[0008] Gene-ECG signal association analysis: Adaptive noise suppression and feature enhancement joint preprocessing operations are performed on the collected gene expression data and ECG signals, and a filtering algorithm that can automatically adjust parameters according to the real-time noise level of the ECG signal is designed. Specifically, the collected ECG signal is short-time Fourier transformed to convert it into the frequency domain, and the energy distribution of the frequency domain signal is simultaneously analyzed to preliminarily determine the type and distribution range of the noise. According to the determined noise type and its distribution, the initial filtering parameters are set for different types of noise. During the filtering process, the noise level of the filtered ECG signal is monitored in real time by calculating the root mean square error of the filtered signal, and the multi-resolution analysis method based on wavelet transform is used to enhance the characteristic information of the R wave. In the association analysis stage, a gene-ECG feature mapping model is constructed, and training is carried out based on the known gene expression and the corresponding ECG signal feature data. The preprocessed gene expression data of the current individual is input into the trained gene-ECG feature mapping model, thereby obtaining the R wave feature prediction result based on gene expression.

[0009] Personalized adjustment of R-wave detection algorithm: Based on the R-wave feature prediction results, the amplitude threshold, slope judgment parameter and adjacent R-wave time interval judgment parameter in the detection algorithm are adjusted accordingly according to the changes in the predicted R-wave amplitude, morphology and occurrence time to adapt to individual differences;

[0010] Adaptive fault-tolerant R-wave detection: The R-wave detection algorithm after personalized adjustment is used for ECG R-wave detection, and an adaptive fault-tolerant mechanism is implemented during the detection process. By comparing with historical data, it monitors whether there are false detections and missed detections. If the threshold is not exceeded, the parameters or algorithms are adjusted in real time according to the current ECG signal characteristics and historical data, and the detection is repeated. If the threshold is exceeded, a prompt message is output to inform that the signal needs to be collected again or the equipment needs to be checked.

[0011] Output and storage of test results: The detected R-wave characteristic information is output for subsequent arrhythmia analysis and cardiovascular disease diagnosis. At the same time, the gene expression data, ECG signals and R-wave test results during this test are stored in the local hard disk and cloud server.

[0012] Furthermore, in the gene expression data and ECG signal collection step, gene expression level information of the individual's entire genome is obtained through gene detection technology, and the gene detection technology includes gene chip technology and high-throughput sequencing technology. When using gene chip technology, the individual's blood sample or tissue sample is processed and hybridized with the gene chip, and the fluorescence signal intensity of each gene site on the chip is detected to obtain gene expression level information. When using high-throughput sequencing technology, the individual's blood sample or tissue sample is subjected to nucleic acid extraction and library construction operations, and then placed in a high-throughput sequencing instrument for sequencing. The gene sequence information obtained by sequencing is processed by bioinformatics analysis methods to calculate the expression level of each gene to form gene expression data.

[0013] Furthermore, in the gene expression data and ECG signal collection step, a multi-channel ECG sensor is used to collect ECG signals, a three-channel or twelve-channel ECG sensor is selected, the number of channels is determined according to the detection needs and accuracy requirements, and the electrode sheets of the ECG sensor are pasted on the ECG electrode placement positions on the individual's chest. The ECG sensor receives the weak electrical signals generated by heart activity through the electrode sheets and converts them into digital signal forms that can be processed by the computer system. During the collection process, the time interval between the collection of gene expression data and ECG signals is ensured to be short to maintain the correlation between the data of the two.

[0014] Furthermore, in the gene-ECG signal association analysis step, a filtering algorithm is designed that can automatically adjust parameters according to the real-time noise level of the ECG signal. The algorithm formula of the filtering algorithm is: Where G(f,t) represents the gain function of the ECG signal after filtering at frequency f and time t, f c (t) is the center frequency reference value that changes dynamically with time t and is used to locate the main noise frequency component. w(t) is the bandwidth parameter that changes dynamically with time t and is used to control the frequency range of the filter. n is the exponential parameter that determines the shape of the filter curve, and its value range is 3≤n≤8. μ is the weight coefficient that balances the noise suppression and R-wave feature enhancement effects, and its value range is 0<μ<1. S represents the number of scale layers of wavelet transform decomposition. ω s is the weighting factor at the sth wavelet scaling level, X s (f) is the frequency domain representation of the ECG signal at the sth wavelet scaling layer, which is obtained by performing wavelet transform on the ECG signal.

[0015] Furthermore, in the gene-ECG signal association analysis step, training is carried out based on known gene expression conditions and corresponding ECG signal feature data, and the model training formula is: Among them, y k represents the predicted value output by the model related to a specific feature of the R wave, σ is an activation function used to introduce nonlinear characteristics so that the model can learn more complex relationships between gene expression and ECG signal features, and w kl is the weight parameter connecting the lth input feature and the kth output feature. The weight is continuously adjusted through the back propagation algorithm during the training process so that the model can accurately learn the potential correlation between gene expression and ECG signal features. k is the bias term corresponding to the kth output feature, which is adjusted during the training process.

[0016] Furthermore, in the adaptive fault-tolerant R-wave detection step, the R-wave detection algorithm is optimized and adjusted in a targeted manner according to the R-wave feature prediction results obtained by the gene-ECG signal correlation analysis. Specifically, when the prediction results show that the amplitude of the R wave changes, when it is predicted that the R-wave amplitude will decrease, the lower amplitude threshold in the R-wave detection algorithm is lowered; when it is predicted that the R-wave amplitude will increase, the upper amplitude threshold in the R-wave detection algorithm is increased; when the prediction results show that the morphology of the R wave will change, when it is predicted that the R-wave morphology becomes flatter, the slope threshold for judging the peak value of the R-wave is lowered; when it is predicted that the R-wave morphology becomes sharper, the slope threshold is increased; when the R-wave appearance time is predicted to be advanced or delayed, when it is predicted that the R-wave appearance time is advanced, the upper limit of the judgment of the time interval between adjacent R-waves is shortened; when it is predicted that the R-wave appearance time is delayed, the lower limit of the judgment of the time interval between adjacent R-waves is extended.

[0017] Furthermore, in the adaptive fault-tolerant R wave detection step, the R wave detection algorithm is optimized and adjusted in a targeted manner according to the R wave feature prediction result obtained by the gene-ECG signal association analysis. Specifically, when the prediction result shows that the amplitude of the R wave changes, the formula is used. The amplitude threshold in the R wave detection algorithm is adjusted, where is the adjusted amplitude threshold, is the original amplitude threshold, α A is the preset amplitude threshold adjustment coefficient, ΔA pred is the change in R wave amplitude predicted by the gene-ECG feature mapping model. If the prediction result shows that the morphology of the R wave will change, the formula is used. Adjust the slope judgment parameters in the R wave detection algorithm, where is the adjusted slope threshold, is the original slope threshold, α S is the preset slope judgment parameter adjustment coefficient, ΔS pred is the slope change corresponding to the R wave morphology change predicted by the gene-ECG feature mapping model. For the predicted advance or delay of the R wave appearance time, the formula is used. The judgment parameters of the time interval between adjacent R waves in the R wave detection algorithm are adjusted, where is the adjusted parameter for determining the time interval between adjacent R waves. is the original adjacent R wave time interval judgment parameter, α I ×ΔT pred is the preset parameter adjustment coefficient for determining the time interval between adjacent R waves, ΔT pred It is the change of R wave appearance time predicted by gene-ECG feature mapping model.

[0018] Furthermore, in the adaptive fault-tolerant R-wave detection step, the formula is used to determine whether false detection and missed detection abnormalities occur by comparing with historical data: in, is the currently detected adjacent R wave spacing value, is the average value of the adjacent R wave spacing in the historical detection data, which is obtained by averaging the adjacent R wave spacing values ​​obtained from multiple previous detections. D is the deviation threshold of the interval between adjacent R waves, and its value range is 0.15≤θ D ≤0.35, is the currently detected R wave amplitude, is the R wave amplitude predicted by the gene-ECG feature mapping model, θ A is the R wave amplitude deviation threshold, the value range is 0.25≤θ A ≤0.45, is the time interval between adjacent R waves currently detected, is the average value of the time interval between adjacent R waves in the historical detection data, θ T is the deviation threshold of the time interval between adjacent R waves, and its value range is 0.1≤θT ≤0.3.

[0019] Furthermore, in the adaptive fault-tolerant R wave detection step, by comparing with historical data to monitor whether there are false detections and missed detections, if the threshold is not exceeded, the parameters or algorithms are adjusted in real time according to the current ECG signal characteristics and historical data and re-detected, and the adjustment formula of the time interval is: in, It is the adjacent R wave time interval judgment parameter currently being used. is the distance between adjacent R waves currently detected The average value of the distance between adjacent R waves in historical detection data The deviation of β I is the adjustment coefficient, the value range is -0.3≤β I ≤0.3, calculate the adjusted adjacent R wave time interval judgment parameter according to the formula Other detection parameters, including amplitude threshold and filtering parameters, are adjusted based on the deviation between the current features and historical data and the set adjustment coefficients to determine the adjusted parameter values, and the adjusted parameters or algorithms are used to re-detect and obtain correct results.

[0020] Compared with the prior art, this human electrocardiogram R wave detection method has the following beneficial effects:

[0021] 1. By introducing gene expression data, the present invention can perform personalized ECG R-wave detection for individuals with specific gene mutations or family-derived cardiovascular disease tendencies. This method not only takes into account the conventional characteristics of the ECG signal itself, but also combines genetic information, thereby being able to more accurately identify R waves and reduce false detections and missed detections. This is of great significance for improving the accuracy of cardiovascular disease diagnosis, early detection of cardiovascular abnormalities, and formulating personalized treatment plans.

[0022] 2. The present invention adopts an adaptive noise suppression and feature enhancement joint preprocessing method, as well as an adaptive fault-tolerant R-wave detection algorithm, to maintain high detection accuracy in complex detection environments and changeable human physiological conditions. By real-time monitoring of the noise level and R-wave characteristics of the ECG signal, and dynamically adjusting the filtering parameters and detection algorithm, it can effectively deal with common interference factors such as electromyographic noise and power frequency interference, while enhancing the characteristic information of the R wave.

[0023] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 It is a flow chart of a method for detecting R-wave of human electrocardiogram;

[0026] Figure 2 The present invention is a flowchart of the steps of gene-ECG signal correlation analysis for a human ECG R wave detection method. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] Embodiment 1

[0029] This embodiment describes in detail a specific application of a human electrocardiogram R-wave detection method in the field of early screening of cardiovascular diseases. By introducing the present invention, more accurate human electrocardiogram R-wave detection can be achieved for individuals with a family history of cardiovascular diseases.

[0030] In the gene expression data and ECG signal collection step, patients undergoing cardiac rehabilitation treatment are selected as monitoring objects, and high-throughput sequencing technology is used to collect the gene expression data of the patients. First, blood samples of the patients are obtained, and the samples are subjected to pre-processing operations such as nucleic acid extraction and library construction to prepare DNA or RNA libraries suitable for sequencing. The constructed libraries are placed in a high-throughput sequencing instrument and sequenced according to the instrument's operating procedures. After sequencing is completed, the gene sequence information obtained by sequencing is processed by bioinformatics analysis methods, and the expression level of each gene is calculated to form the patient's gene expression data. At the same time, a three-channel ECG sensor is used to collect ECG signals from the same patient, and electrode sheets are attached to the patient's chest according to the standard ECG electrode placement position to ensure that the electrode sheets are in good contact with the skin and reduce signal interference. ECG signals for a period of time (such as 10 minutes) are collected, and the collected gene expression data and ECG signals are transmitted to a computer system through a data cable for subsequent processing.

[0031] In the gene-ECG signal association analysis step, the collected ECG signals are preprocessed by adaptive noise suppression and feature enhancement. First, the designed filtering algorithm is used Perform short-time Fourier transform on the ECG signal, convert it to the frequency domain, analyze the spectrum characteristics and energy distribution, and find that there is a certain degree of power frequency interference and myoelectric noise. For power frequency interference, set the f in the filtering algorithm according to the power frequency of the area (such as 50Hz). c (t) is close to 50Hz, and the appropriate B is determined according to the actual noise situation. w (t), for myoelectric noise, the corresponding f is determined by analyzing its energy distribution c (t) and B w (t), during the filtering process, the root mean square error (RMSE) of the filtered signal is calculated to monitor the noise level in real time. If the RMSE value is high, the parameter adjustment rule in the formula is used to appropriately increase B. w (t) value to further suppress noise. At the same time, the multi-resolution analysis method based on wavelet transform is used to enhance the R wave feature. The ECG signal is subjected to wavelet transform and decomposed into different scales (such as setting S = 5 scale layers). The frequency domain representation X of the ECG signal at each scale is obtained. s (f) Based on the characteristic performance of R waves at different scales obtained from the previous analysis of a large number of normal and abnormal ECG signals, the weighting factor ω at each scale layer is determined. s , for the scale layer where the R wave peak is located, set a relatively large ω s Value, according to the formula The decomposed coefficients are weighted to make the R wave more prominent and complete in the reconstructed signal. In the association analysis stage, a gene-ECG feature mapping model is constructed, and a convolutional neural network (CNN) is selected as the training algorithm. A large number of gene expression conditions and corresponding ECG signal feature data of individuals with a family history of cardiovascular disease are collected as training sets. The input features of the model are determined to include the expression level values ​​of each gene in the preprocessed gene expression data and the relevant feature values ​​in the ECG signal feature data (such as the amplitude, slope, and time interval of adjacent R waves of the R wave). The output is a predicted value related to a specific feature of the R wave (such as the predicted R wave amplitude change ratio, the degree of morphological change, the advance or delay of the appearance time, etc.). According to the formula The model is trained, where the activation function σ is RelU. During the training process, the mean square error loss function is minimized. (N is the number of training samples, y k,i is the predicted value of the i-th training sample, is the actual observed value of the eighth training sample) to adjust the weight parameter w kl and the bias term b k, so that the model is continuously optimized, the pre-processed gene expression data of the current individual is input into the trained gene-ECG feature mapping model, and the R-wave feature prediction results based on gene expression are obtained. For example, it is predicted that the R-wave amplitude in the ECG signal of the individual may be reduced by a certain proportion, and the R-wave morphology may become flatter.

[0032] In the step of personalized adjustment of the R wave detection algorithm, the traditional R wave detection algorithm is personalized adjusted according to the above prediction results. Since it is predicted that the R wave amplitude may decrease, the formula is used. Adjust the amplitude threshold in the R wave detection algorithm. The original amplitude threshold in the traditional R wave detection algorithm is known. The average amplitude reference value A0 of the R wave under normal conditions is obtained by statistically analyzing the R wave amplitude in a large number of normal ECG signal samples. The R wave amplitude change ΔA predicted by the gene-ECG feature mapping model is pred (Assuming that the R wave amplitude is predicted to decrease by 20%, ΔA pred =-0.2A0), set the amplitude threshold adjustment coefficient α A =-0.5, and the adjusted R wave detection amplitude threshold is calculated according to the formula The lower amplitude threshold in the R wave detection algorithm is lowered accordingly, because the R wave shape is predicted to become flatter, using the formula Adjust the slope judgment parameters in the R wave detection algorithm. The original slope threshold in the traditional R wave detection algorithm is known. The slope change ΔS corresponding to the R wave morphology change predicted by the gene-ECG feature mapping model pred (Assuming that the R wave shape is predicted to become flatter and the slope will decrease by a certain value, then ΔS pred is the corresponding negative slope change), and the slope judgment parameter adjustment coefficient α is set S =-0.4, and the adjusted slope threshold for determining the peak value of the R wave is calculated according to the formula and lower the slope threshold used to determine the peak of the R wave.

[0033] In the adaptive fault-tolerant R wave detection step, the detection parameters and related variables are initialized, the false detection threshold is set to 3, and the missed detection threshold is set to 5. According to the personalized R wave detection algorithm, the R wave is searched in the preprocessed ECG signal, and the characteristic information such as the position, amplitude and shape of the R wave is identified. The results of each detection and related intermediate data, such as the spacing between adjacent R waves and the stability index of the overall signal, are recorded. During the detection process, whether there are abnormal situations such as false detection and missed detection in real time is monitored, and the formula is used. Make a judgment, among which, is the currently detected adjacent R wave spacing value, is the average value of the distance between adjacent R waves in the historical detection data, and θ D =0.25, is the currently detected R wave amplitude, is the R wave amplitude predicted by the gene-ECG feature mapping model, let θ A =0.35, is the time interval between adjacent R waves currently detected, is the average value of the time interval between adjacent R waves in the historical detection data, and θ T = 0.2, if an abnormality is detected and the number of false detections and missed detections does not exceed the set threshold, for example, if the relative deviation between the interval between adjacent R waves and the historical average interval exceeds θ D Taking the adjustment of the time interval judgment parameter of adjacent R waves as an example, the formula is used Adjust the adjacent R wave time interval judgment parameters currently in use. Assume adjustment coefficient β I =-0.2, and the adjusted adjacent R wave time interval judgment parameter is calculated according to the formula based on the deviation between the currently detected adjacent R wave interval and the historical average interval Then use the adjusted parameters or algorithms to re-test and try to obtain the correct result.

[0034] In the step of outputting and storing the test results, the detected R-wave position, amplitude, morphology and other characteristic information are output in the form of a visual chart through the display screen. The position information is displayed with the time axis as the horizontal axis and the ECG signal amplitude as the vertical axis. The amplitude information is marked at the corresponding position in the form of a specific numerical value. The morphological information is presented by drawing the R-wave curve and marking the key characteristic points. The gene expression data, ECG signal and R-wave test results in this test process are stored in the local hard disk and classified and labeled according to the test time, individual identification and other information for subsequent reference and further analysis.

[0035] In summary, the present invention can achieve more accurate human ECG R-wave detection in the field of early screening of cardiovascular diseases for individuals with a family history of cardiovascular diseases, which helps to discover potential abnormal heart rhythms and signs of cardiovascular diseases earlier, and provide strong support for subsequent diagnosis and treatment.

[0036] Embodiment 2

[0037] This embodiment describes in detail a specific application of a human ECG R-wave detection method in the field of personalized cardiac rehabilitation monitoring. By introducing the present invention, accurate detection and dynamic monitoring of R-waves are achieved based on the patient's individual gene expression and real-time ECG signals, which helps rehabilitation therapists and doctors to more accurately grasp the changes in the patient's heart condition and adjust the rehabilitation plan in time.

[0038] In the gene expression data and ECG signal collection step, patients undergoing cardiac rehabilitation treatment are selected as monitoring objects, and high-throughput sequencing technology is used to collect the gene expression data of the patients. First, blood samples of the patients are obtained, and the samples are subjected to pre-processing operations such as nucleic acid extraction and library construction to prepare a DNA or RNA library suitable for sequencing. The constructed library is placed in a high-throughput sequencing instrument and sequenced according to the instrument's operating procedures. After sequencing is completed, the gene sequence information obtained by sequencing is processed by bioinformatics analysis methods, and the expression level of each gene is calculated to form the patient's gene expression data. At the same time, a three-channel ECG sensor is used to collect the ECG signal of the same patient, and an electrode sheet is attached to the patient's chest according to the standard ECG electrode placement position to ensure that the electrode sheet is in good contact with the skin and reduce signal interference. ECG signals for a period of time (such as 10 minutes) are collected, and the collected gene expression data and ECG signals are transmitted to a computer system through a data cable for subsequent processing.

[0039] In the gene-ECG signal association analysis step, the collected ECG signals are preprocessed by adaptive noise suppression and feature enhancement. The designed filtering algorithm is used to perform short-time Fourier transform on the ECG signals, convert them into the frequency domain, analyze the spectrum characteristics and energy distribution, and find that there is a small amount of power frequency interference and relatively obvious myoelectric noise. For power frequency interference, the f in the filtering algorithm is set to c (t) Close to the power frequency of the area (such as 60Hz), and determine the appropriate B according to the actual noise situation. w (t), for myoelectric noise, the corresponding f is determined by analyzing its energy distribution c (t) and B R (t), during the filtering process, the signal-to-noise ratio (SNR) of the filtered signal is calculated to monitor the noise level in real time. If the SNR value is found to be low, B is appropriately increased according to the parameter adjustment rules in the formula. w (t) value to further suppress noise. At the same time, the multi-resolution analysis method based on wavelet transform is used to enhance the R wave feature. The ECG signal is subjected to wavelet transform and decomposed into different scales (such as setting S = 4 scale layers). The frequency domain representation X of the ECG signal at each scale is obtained. s (f) Based on the characteristic performance of R waves at different scales obtained from the previous analysis of a large number of normal and abnormal ECG signals, the weighting factor ω at each scale layer is determined. s , for the scale layer where the R wave peak is located, set a relatively large ω s Value, according to the formula The decomposed coefficients are weighted to make the R wave more prominent and complete in the reconstructed signal. In the association analysis stage, a gene-ECG feature mapping model is constructed, and the support vector machine (SVM) is selected as the training algorithm. The radial basis function (RBF) is selected as the kernel function. A large number of gene expression conditions and corresponding ECG signal feature data of patients undergoing cardiac rehabilitation treatment are collected as training sets. The input features of the model are determined to include the expression level values ​​of each gene in the preprocessed gene expression data and the relevant feature values ​​in the ECG signal feature data (such as the amplitude, slope, and time interval of adjacent R waves of the R wave). The output is a predicted value related to a specific feature of the R wave (such as the predicted R wave amplitude change ratio, the degree of morphological change, the advance or delay of the appearance time, etc.). According to the formula Model training is performed, where the activation function f determines its specific form according to the selected algorithm. During the training process, the mean square error loss function is minimized. (N is the number of training samples, y k,i is the predicted value of the i-th training sample, Adjust the weight parameter w for the actual observation value of the i-th training sample kl and the bias term b k , so that the model is continuously optimized, the pre-processed gene expression data of the current patient is input into the trained gene-ECG feature mapping model, and the R-wave feature prediction results based on gene expression are obtained. For example, it is predicted that the appearance time of the R wave in the patient's ECG signal may be delayed by a certain amount, and the R-wave amplitude may increase by a certain proportion.

[0040] In the step of personalized adjustment of the R wave detection algorithm, the traditional R wave detection algorithm is personalized adjusted according to the above prediction results. Since it is predicted that the R wave amplitude may increase, the formula is used. Adjust the amplitude threshold in the R wave detection algorithm. The original amplitude threshold in the traditional R wave detection algorithm is known. The average amplitude reference value A0 of the R wave under normal conditions is obtained by statistically analyzing the R wave amplitude in a large number of normal ECG signal samples. The R wave amplitude change ΔA predicted by the gene-1 ECG feature mapping model is pred (Assuming that the R wave amplitude is predicted to increase by 3096, then ΔA pred =0.3A0), set the amplitude threshold adjustment coefficient α A =0.4, and the adjusted R wave detection amplitude threshold is calculated according to the formula Accordingly, the upper threshold of the amplitude in the R wave detection algorithm is increased because the predicted R wave appearance time may be delayed. The formula is used The judgment parameters of the time interval between adjacent F waves in the R wave detection algorithm are adjusted. The original judgment parameters of the time interval between adjacent R waves in the traditional R wave detection algorithm are known. By counting the time intervals of adjacent R waves in a large number of normal ECG signal samples, the average time interval reference value T0 of adjacent R waves under normal circumstances is obtained, and the R wave appearance time change ΔT predicted by the gene-ECG feature mapping model is pred (Assuming that the predicted R wave is delayed by 0.2 seconds, ΔT pred =0.2T0), let the adjacent R wave time interval judgment parameter adjustment coefficient α I =0.3, and the adjusted adjacent R wave time interval judgment parameter is calculated according to the formula.

[0041] In the adaptive fault-tolerant R wave detection, the detection parameters and related variables are initialized, the false detection threshold is set to 2, the missed detection threshold is set to 4, and the R wave is searched in the preprocessed ECG signal according to the personalized R wave detection algorithm. The position, amplitude and shape of the R wave and other characteristic information are identified, and the results of each detection and related intermediate data, such as the spacing between adjacent R waves and the stability index of the overall signal, are recorded. During the detection process, whether there are abnormal situations such as false shooting and missed detection in real time is monitored, and the formula is used. Make a judgment, among which, is the adjacent spacing value of the adjacent R waves currently detected, is the average value of the distance between adjacent R waves in the historical detection data, and θ D =0.2; is the currently detected R wave amplitude, is the R wave amplitude predicted by the gene-ECG feature mapping model, let θ A =0.3; is the time interval between adjacent R waves currently detected, is the average value of the time interval between adjacent R waves in the historical detection data, and θ T =0.15, if an abnormality is detected and the number of false detections and missed detections does not exceed the set threshold, for example, if the relative deviation between the interval between adjacent R waves and the historical average interval exceeds θ D Taking the adjustment of the time interval judgment parameter of adjacent R waves as an example, the formula is used Adjust the adjacent R wave time interval judgment parameters currently in use. Assume adjustment coefficient β I =-0.25, and the adjusted adjacent R wave time interval judgment parameter is calculated according to the formula based on the relative deviation between the currently detected adjacent R wave interval and the historical average interval Then use the adjusted parameters or algorithms to re-test and try to obtain the correct result.

[0042] In the step of outputting and storing the test results, the detected R-wave position, amplitude, morphology and other characteristic information are output in the form of a visual chart through the display screen. The position information is displayed with the time axis as the horizontal axis and the ECG signal amplitude as the vertical axis. The amplitude information is marked at the corresponding position in the form of a specific numerical value. The morphological information is presented by drawing the R-wave curve and marking the key characteristic points. The rehabilitation treatment team (including doctors, rehabilitation therapists, etc.) can intuitively view the relevant characteristics of the R-wave in order to accurately evaluate the changes in the patient's cardiac electrophysiological state during the cardiac rehabilitation process, and then adjust the rehabilitation plan. The gene expression data, ECG signals and R-wave test results in this test process are stored in the cloud server, and are classified and labeled according to the test time, patient identification and other information for subsequent reference and further analysis. Cloud storage facilitates professionals in different locations to access data at any time, conduct remote collaborative diagnosis or track the patient's rehabilitation progress.

[0043] In summary, in the field of personalized cardiac rehabilitation monitoring, the present invention can achieve accurate human ECG R-wave detection based on the patient's individual gene expression and real-time ECG signals, and through adaptive adjustment and monitoring mechanisms, timely capture changes in R-wave characteristics, and provide more targeted monitoring and evaluation methods for cardiac rehabilitation treatment, which helps to optimize rehabilitation plans, improve rehabilitation effects, and ensure the patient's heart health recovery.

[0044] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for detecting human electrocardiogram R wave, characterized in that: The method comprises the following specific steps: Gene expression data and ECG signal collection: Gene expression level information of the whole genome of an individual is obtained through gene detection technology, and ECG signals are collected using multi-channel ECG sensors. The time interval between the two collections is kept short, and the collected gene expression data and ECG signals are transmitted to the computer system; Gene-ECG signal association analysis: Adaptive noise suppression and feature enhancement joint preprocessing operations are performed on the collected gene expression data and ECG signals, and a filtering algorithm that can automatically adjust parameters according to the real-time noise level of the ECG signal is designed. Specifically, the collected ECG signal is short-time Fourier transformed to convert it into the frequency domain, and the energy distribution of the frequency domain signal is simultaneously analyzed to preliminarily determine the type and distribution range of the noise. According to the determined noise type and its distribution, the initial filtering parameters are set for different types of noise. During the filtering process, the noise level of the filtered ECG signal is monitored in real time by calculating the root mean square error of the filtered signal, and the multi-resolution analysis method based on wavelet transform is used to enhance the characteristic information of the R wave. In the association analysis stage, a gene-ECG feature mapping model is constructed, and training is carried out based on the known gene expression and the corresponding ECG signal feature data. The preprocessed gene expression data of the current individual is input into the trained gene-ECG feature mapping model, thereby obtaining the R wave feature prediction result based on gene expression. Personalized adjustment of R-wave detection algorithm: Based on the R-wave feature prediction results, the amplitude threshold, slope judgment parameter and adjacent R-wave time interval judgment parameter in the detection algorithm are adjusted accordingly according to the changes in the predicted R-wave amplitude, morphology and occurrence time to adapt to individual differences; Adaptive fault-tolerant R-wave detection: The R-wave detection algorithm after personalized adjustment is used for ECG R-wave detection, and an adaptive fault-tolerant mechanism is implemented during the detection process. By comparing with historical data, it monitors whether there are false detections and missed detections. If the threshold is not exceeded, the parameters or algorithms are adjusted in real time according to the current ECG signal characteristics and historical data, and the detection is repeated. If the threshold is exceeded, a prompt message is output to inform that the signal needs to be collected again or the equipment needs to be checked. Output and storage of test results: The detected R-wave characteristic information is output for subsequent arrhythmia analysis and cardiovascular disease diagnosis. At the same time, the gene expression data, ECG signals and R-wave test results during this test are stored in the local hard disk and cloud server.

2. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the gene expression data and ECG signal acquisition step, the gene expression level information of the individual's whole genome is obtained by gene detection technology, and the gene detection technology includes gene chip technology and high-throughput sequencing technology. When using gene chip technology, the individual's blood sample or tissue sample is processed and hybridized with the gene chip, and the fluorescence signal intensity of each gene site on the chip is detected to obtain the gene expression level information. When using high-throughput sequencing technology, the individual's blood sample or tissue sample is subjected to nucleic acid extraction and library construction operations, and then placed in a high-throughput sequencing instrument for sequencing. The gene sequence information obtained by sequencing is processed by a bioinformatics analysis method to calculate the expression level of each gene to form gene expression data.

3. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the gene expression data and ECG signal collection step, a multi-channel ECG sensor is used to collect ECG signals. A three-channel or twelve-channel ECG sensor is selected, and the number of channels is determined according to the detection needs and accuracy requirements. The electrode sheets of the ECG sensor are pasted on the ECG electrode placement positions on the individual's chest. The ECG sensor receives the weak electrical signals generated by heart activity through the electrode sheets and converts them into digital signal forms that can be processed by a computer system. During the collection process, the time interval between the collection of gene expression data and ECG signals is ensured to be short to maintain the correlation between the data of the two.

4. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the gene-ECG signal association analysis step, a filtering algorithm is designed that can automatically adjust parameters according to the real-time noise level of the ECG signal. The algorithm formula of the filtering algorithm is: Where G(f,t) represents the gain function of the ECG signal after filtering at frequency f and time t, f c (t) is the center frequency reference value that changes dynamically with time t and is used to locate the main noise frequency component. w (t) is the bandwidth parameter that changes dynamically with time t and is used to control the frequency range of the filter. n is the exponential parameter that determines the shape of the filter curve, and its value range is 3≤n≤8. μ is the weight coefficient that balances the noise suppression and R-wave feature enhancement effects, and its value range is 0<μ<1. S represents the number of scale layers of wavelet transform decomposition. ω s is the weighting factor at the sth wavelet scaling level, X s (f) is the frequency domain representation of the ECG signal at the sth wavelet scaling layer, which is obtained by performing wavelet transform on the ECG signal.

5. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the gene-ECG signal association analysis step, training is carried out based on known gene expression conditions and corresponding ECG signal feature data, and the model training formula is: Among them, y k represents the predicted value output by the model related to a specific feature of the R wave, σ is an activation function used to introduce nonlinear characteristics so that the model can learn more complex relationships between gene expression and ECG signal features, and w kl is the weight parameter connecting the lth input feature and the kth output feature. The weight is continuously adjusted through the back propagation algorithm during the training process so that the model can accurately learn the potential correlation between gene expression and ECG signal features. k is the bias term corresponding to the kth output feature, which is adjusted during the training process.

6. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the adaptive fault-tolerant R-wave detection step, the R-wave detection algorithm is optimized and adjusted in a targeted manner according to the R-wave feature prediction results obtained by the gene-ECG signal association analysis. Specifically, when the prediction results show that the amplitude of the R wave changes, when it is predicted that the R-wave amplitude will decrease, the lower amplitude threshold in the R-wave detection algorithm is lowered; when it is predicted that the R-wave amplitude will increase, the upper amplitude threshold in the R-wave detection algorithm is increased; when the prediction results show that the morphology of the R wave will change, when it is predicted that the R-wave morphology becomes flatter, the slope threshold for judging the peak value of the R-wave is lowered; when it is predicted that the R-wave morphology becomes sharper, the slope threshold is increased; when the occurrence time of the R wave is predicted to be advanced or delayed, when it is predicted that the occurrence time of the R wave is advanced, the upper limit of the judgment of the time interval between adjacent R waves is shortened; when it is predicted that the occurrence time of the R wave is delayed, the lower limit of the judgment of the time interval between adjacent R waves is extended.

7. A human electrocardiogram R wave detection method according to claim 6, characterized in that: In the adaptive fault-tolerant R wave detection step, the R wave detection algorithm is optimized and adjusted in a targeted manner according to the R wave feature prediction results obtained by the gene-ECG signal association analysis. Specifically, when the prediction results show that the amplitude of the R wave changes, the formula is used. The amplitude threshold in the R wave detection algorithm is adjusted, where is the adjusted amplitude threshold, is the original amplitude threshold, α A is the preset amplitude threshold adjustment coefficient, ΔA pred is the change in R wave amplitude predicted by the gene-ECG feature mapping model. If the prediction result shows that the morphology of the R wave will change, the formula is used. Adjust the slope judgment parameters in the R wave detection algorithm, where is the adjusted slope threshold, is the original slope threshold, α S is the preset slope judgment parameter adjustment coefficient, ΔS pred is the slope change corresponding to the R wave morphology change predicted by the gene-ECG feature mapping model. For the predicted advance or delay of the R wave appearance time, the formula is used. The judgment parameters of the time interval between adjacent R waves in the R wave detection algorithm are adjusted, where is the adjusted parameter for determining the time interval between adjacent R waves. is the original adjacent R wave time interval judgment parameter, α I ×ΔT pred is the preset parameter adjustment coefficient for determining the time interval between adjacent R waves, ΔT pred It is the change of R wave appearance time predicted by gene-ECG feature mapping model.

8. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the R wave detection step of the adaptive fault tolerance, the formula is used to judge whether there is a false detection or missed detection abnormality by comparing with the historical data: in, is the currently detected adjacent R wave spacing value, is the average value of the adjacent R wave spacing in the historical detection data, which is obtained by averaging the adjacent R wave spacing values ​​obtained from multiple previous detections. D is the deviation threshold of the interval between adjacent R waves, and its value range is 0.15≤θ D ≤0.35, is the currently detected R wave amplitude, is the R wave amplitude predicted by the gene-ECG feature mapping model, θ A is the R wave amplitude deviation threshold, the value range is 0.25≤θ A ≤0.45, is the time interval between adjacent R waves currently detected, is the average value of the time interval between adjacent R waves in the historical detection data, θ T is the deviation threshold of the time interval between adjacent R waves, and its value range is 0.1≤θ T ≤0.

3.

9. A human electrocardiogram R wave detection method according to claim 1, characterized in that: In the R wave detection step of the adaptive fault-tolerant method, by comparing with the historical data, whether there is any false detection or missed detection abnormality, if it does not exceed the threshold, the parameters or algorithm are adjusted in real time according to the current ECG signal characteristics and historical data and re-detected, and the adjustment formula of the time interval is: in, It is the adjacent R wave time interval judgment parameter currently being used. is the distance between adjacent R waves currently detected The average value of the distance between adjacent R waves in historical detection data The deviation of β I is the adjustment coefficient, the value range is -0.3≤β I ≤0.3, calculate the adjusted adjacent R wave time interval judgment parameter according to the formula Other detection parameters, including amplitude threshold and filtering parameters, are adjusted based on the deviation between the current features and historical data and the set adjustment coefficients to determine the adjusted parameter values, and the adjusted parameters or algorithms are used to re-detect and obtain correct results.

Images