Method, system and storage medium for generating electrocardiogram data

Abstract

The present invention relates to the field of electrocardiogram technology, and in particular to a method, system and storage medium for generating electrocardiogram data. The method comprises the following steps: collecting user electrocardiogram signals through electrodes and performing fluctuation morphology analysis to obtain electrocardiogram fluctuation morphology data; performing abnormal electrocardiogram fluctuation identification on the electrocardiogram fluctuation morphology data, and then performing cluster distribution density calculation to obtain abnormal cluster distribution density data; performing sample feature sampling based on the abnormal cluster distribution density data to obtain abnormal evolution trend sample data; performing learning iteration on the abnormal evolution trend sample data to obtain abnormal evolution trend learning data. The present invention makes the electrocardiogram technology more perfect by optimizing the electrocardiogram technology.

Description

Technical Field

[0001] The present invention relates to the field of electrocardiogram technology, and in particular to a method, system and storage medium for generating electrocardiogram data. Background Art

[0002] Electrocardiogram (ECG) is a graphical representation of the heart's electrical activity, which is widely used in the diagnosis, monitoring and prevention of cardiovascular diseases. ECG recording relies on physical equipment, such as electrocardiographs, patch sensors, etc., which detect changes in the heart's electrical signals on the skin surface to generate waveforms. The generation of synthetic ECG data has become a research hotspot in recent years. This method is based on the modeling of the human heart's electrical activity, and uses mathematical models or machine learning methods to generate waveform data similar to the real ECG. Synthetic ECG can solve the noise problem existing in the acquisition process of traditional equipment, and at the same time provide an ECG data source that can be stably output under various conditions. For example, by using mathematical modeling technology, ECG waveforms under different heart disease states can be simulated, thereby providing a large amount of simulated data for medical research. In addition, data-driven methods rely on large-scale real ECG data to generate synthetic data through deep learning and other means. These data have shown high application value in the study of pathological conditions such as arrhythmia and myocardial ischemia. However, the traditional method of generating ECG data has the problem of inaccurate recognition of abnormal ECG signals and inaccurate judgment of the evolution trend of ECG signals. Summary of the invention

[0003] Based on this, it is necessary to provide a method, system and storage medium for generating electrocardiogram data to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for generating electrocardiogram data comprises the following steps:

[0005] Step S1: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal; performing a fluctuation morphology analysis on the user's ECG signal to obtain ECG fluctuation morphology data;

[0006] Step S2: performing abnormal ECG fluctuation recognition on the ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; performing cluster distribution density calculation on the abnormal ECG fluctuation data to obtain abnormal cluster distribution density data;

[0007] Step S3: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain the abnormal ECG evolution trend data; sampling the sample features of the abnormal ECG evolution trend data to obtain the abnormal evolution trend sample data;

[0008] Step S4: Perform learning iteration on the abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; perform abnormal ECG labeling processing on the ECG fluctuation morphology data based on the abnormal evolution trend learning data, and feed it back to the terminal to execute the generation of ECG data.

[0009] The present invention collects the user's ECG signal through an electrode sheet, which is the basis of electrocardiogram (ECG) monitoring. The electrode sheet can effectively capture small changes in the heart's electrical activity and convert it into an electrical signal for analysis. Subsequently, by performing a wave morphology analysis on the collected ECG signal, key features in the ECG, such as heartbeat intervals, peaks and troughs, can be extracted. This process can not only reveal the health status of the user's heart, but also provide a data basis for subsequent abnormal identification. By analyzing the ECG wave morphology data, the characteristics of normal ECG signals can be identified, and standards can be set for the detection of abnormal signals, thereby improving the accuracy and effectiveness of subsequent analysis. Abnormal ECG fluctuations are identified on the ECG wave morphology data obtained by step S1. This process uses advanced algorithms, such as machine learning and statistical analysis, to detect abnormal patterns in ECG waveforms. These abnormal fluctuations are indicators of potential health problems such as arrhythmias and myocardial ischemia. Subsequently, the cluster distribution density calculation of the identified abnormal ECG fluctuation data can help further analyze the distribution and density of these abnormal fluctuations. Through this calculation, the frequency of occurrence, concentrated areas and clinical significance of abnormal fluctuations can be identified. This not only helps to improve the detection rate of abnormal fluctuations, but also provides an important basis for clinical decision-making. According to the abnormal cluster distribution density data, the abnormal ECG fluctuation data is predicted for abnormal evolution trend. This process uses time series analysis and prediction models to identify the future development trend of abnormal fluctuations. A better understanding of the evolution of the patient's ECG state can also provide a basis for early intervention. Subsequently, the predicted ECG abnormal evolution trend data is sampled for sample features to extract representative feature samples. By extracting sample features, the accuracy and effectiveness of subsequent analysis can be ensured, laying the foundation for in-depth research on abnormal evolution trends. The realization of this process enhances the dynamic monitoring ability of heart health status, provides a scientific basis for early identification of potential cardiovascular diseases, and helps medical personnel make decisions more scientifically and efficiently. The random forest algorithm is used to learn and iterate the abnormal evolution trend sample data. With its powerful classification and regression capabilities, this algorithm shows excellent results when processing complex ECG fluctuation data. Through continuous learning and optimization, random forest can improve the recognition accuracy of abnormal ECG fluctuations and establish a more accurate abnormal evolution trend model. Based on these learning data, the system will perform abnormal ECG labeling on ECG wave morphology data. This labeling result will be fed back to the terminal to generate ECG data, thereby providing an intuitive and easy-to-understand clinical data display. This technical solution effectively integrates data collection, analysis and feedback, ensuring the timeliness and accuracy of information transmission, not only improving the intelligent level of ECG monitoring, but also providing a powerful decision support tool to promote the early detection and intervention of cardiovascular diseases.Therefore, the present invention is an improvement on a traditional method for generating electrocardiogram data, which solves the problems of inaccurate recognition of abnormal electrocardiogram signals and inability to accurately judge the evolution trend of electrocardiogram signals in the traditional method for generating electrocardiogram data, improves the accuracy of identifying abnormal electrocardiogram signals and improves the ability to accurately judge the evolution trend of electrocardiogram signals.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal;

[0012] Step S12: performing signal denoising processing on the user's ECG signal to obtain a user's ECG denoised signal;

[0013] Step S13: performing baseline drift removal processing on the user's ECG signal according to the user's ECG denoising signal to obtain the user's ECG baseline drift removal data;

[0014] Step S14: performing fluctuation shape analysis on the user's ECG baseline drift-free data to obtain ECG fluctuation shape data.

[0015] The present invention collects the user's ECG signal through the electrode sheet. This process is the basis of ECG monitoring, and the placement and number of the electrode sheets will directly affect the quality and accuracy of the signal. The electrode can effectively capture the electrical activity of the heart and convert these weak electrical signals into digital signals for subsequent analysis. Accurate signal acquisition not only provides a solid foundation for subsequent data processing, but also can timely reflect the user's heart health status. Through the collected original ECG signal, the subsequent data cleaning and analysis steps will be able to work more effectively, thereby improving the effectiveness and reliability of the overall monitoring. The user's ECG signal is subjected to signal denoising to improve the signal quality. ECG signals are often interfered by various noises, such as myoelectric noise, motion artifacts and power supply interference. These noises will seriously affect the accuracy of the signal, resulting in unreliable subsequent analysis results. By adopting appropriate denoising algorithms, such as wavelet transform or filtering technology, these interferences can be effectively eliminated and the main features of the ECG signal can be retained. The denoised signal will be smoother and clearer, providing a reliable data basis for subsequent baseline drift processing and fluctuation morphology analysis, thereby improving the accuracy and reliability of subsequent analysis. Baseline drift removal is performed based on the user's ECG denoising signal. This processing link is one of the key steps in ECG analysis, because baseline drift is often caused by various factors, such as poor electrode contact, movement or breathing. By applying a suitable baseline drift removal algorithm, the changing part of the baseline can be removed to obtain more realistic ECG fluctuation information. This process ensures that the subsequent fluctuation morphology analysis will not be affected by baseline drift, thereby more accurately reflecting the electrical activity of the heart. The baseline drift-free data provides a more reliable basis for the subsequent fluctuation morphology analysis, making the ECG interpretation more accurate. The user's ECG baseline drift-free data is subjected to fluctuation morphology analysis. This analysis aims to extract key features in ECG signals, such as the morphology, amplitude and interval of P waves, QRS waves and T waves. These fluctuation morphologies can not only reflect the electrophysiological characteristics of the heart, but also reveal potential heart health problems. By using signal processing algorithms and pattern recognition technology, ECG fluctuation morphologies can be deeply quantitatively analyzed to obtain ECG fluctuation morphology data. The implementation of this step greatly improves the ability to interpret ECG signals and provides important data information for subsequent abnormality identification and trend prediction.

[0016] Preferably, step S2 comprises the following steps:

[0017] Step S21: constructing a Cartesian coordinate system for the ECG wave shape data to obtain an ECG wave shape coordinate system;

[0018] Step S22: performing abnormal ECG fluctuation identification on the ECG fluctuation morphology data according to the ECG fluctuation morphology coordinate system to obtain abnormal ECG fluctuation data;

[0019] Step S23: performing abnormal clustering analysis on the abnormal ECG fluctuation data to obtain abnormal ECG clustering data;

[0020] Step S24: Calculate the cluster distribution density of the abnormal ECG fluctuation data according to the abnormal ECG cluster data to obtain abnormal cluster distribution density data.

[0021] The present invention presents the fluctuation information of the ECG signal in a visual form by constructing a Cartesian coordinate system for the ECG fluctuation morphology data. This coordinate system provides a standardized way to map different characteristic points (such as peaks and troughs) of the ECG signal into a two-dimensional space, which is convenient for subsequent data analysis and processing. Through this construction, the time and amplitude characteristics of the signal can be more intuitively displayed, which is convenient for observing and analyzing the changing trend of the ECG fluctuation. Using the Cartesian coordinate system, researchers and medical professionals can better understand and interpret the characteristics of ECG fluctuations, thereby providing a scientific basis for abnormal identification and diagnosis. Based on the ECG fluctuation morphology coordinate system, the ECG fluctuation morphology data is identified for abnormal ECG fluctuations. This process mainly relies on advanced algorithms, such as machine learning and statistical methods, to detect abnormal patterns in the signal. By comparing the standard ECG fluctuation characteristics, the system can identify abnormal fluctuation data that exceeds the normal range. This identification not only helps to discover potential heart health problems, but also provides early warnings for patients and promotes timely medical intervention. Accurate abnormal identification capability is the core of the entire ECG monitoring system, laying the foundation for subsequent abnormal cluster analysis and trend prediction, and ensuring the efficiency and reliability of patient health management. Abnormal clustering analysis is performed on the identified abnormal ECG fluctuation data. This analysis aims to evaluate the distribution characteristics of abnormal fluctuations in time and space and identify whether there is a clustering phenomenon of abnormal ECG fluctuations. By using statistical analysis methods and clustering algorithms, it is possible to determine whether these abnormal fluctuations occur frequently in a specific period of time or under specific conditions, revealing their potential pathological significance. This clustering analysis can not only help medical professionals understand the mechanism of abnormal fluctuations, but also provide an important basis for subsequent intervention measures. By deeply analyzing the clustering of abnormal fluctuations, the risk of heart disease can be more accurately identified, and corresponding treatment plans can be formulated in a targeted manner. Based on the ECG abnormal clustering data, the clustering distribution density of abnormal ECG fluctuation data is calculated. This process uses mathematical statistical methods to evaluate the distribution density of abnormal ECG fluctuations in different time periods and under different conditions. This density calculation can reveal the frequency and distribution characteristics of abnormal fluctuations, helping medical professionals identify high-incidence periods and high-risk factors of ECG abnormalities. Through this calculation, personalized health management suggestions can be provided to patients to promote early intervention and monitoring of diseases. At the same time, the data also provides valuable reference information for medical research, helps to explore the relationship between abnormal ECG fluctuations and other clinical parameters, and promotes scientific research on cardiovascular diseases.

[0022] Preferably, step S22 comprises the following steps:

[0023] Step S221: dividing the ECG fluctuation morphology data into time series fluctuation intervals according to the ECG fluctuation morphology coordinate system to obtain the ECG fluctuation time series fluctuation intervals;

[0024] Step S222: calculating the mean difference of the fluctuation rising rates of the ECG fluctuation morphology data in different fluctuation intervals based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system to obtain the fluctuation rising rate mean difference data;

[0025] Step S223: Capture the multi-scale dynamic changes of the fluctuation rising rate mean difference data to obtain the multi-scale dynamic change data of the rising rate;

[0026] Step S224: calculating the regional decline rate ratios of different fluctuation intervals of the ECG fluctuation morphology data based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system to obtain the ECG fluctuation regional decline rate ratios;

[0027] Step S225: identifying the descending acceleration regions of different fluctuation intervals of the ECG fluctuation shape data according to the ECG fluctuation region descending rate ratio, and obtaining ECG descending acceleration region data;

[0028] Step S226: Identify abnormal ECG fluctuations based on the multi-scale dynamic change data of the rising rate, the ECG fluctuation area falling rate ratio and the ECG falling acceleration area data to obtain abnormal ECG fluctuation data.

[0029] The present invention divides the ECG fluctuation morphology data into time series fluctuation intervals according to the ECG fluctuation morphology coordinate system. This process is to divide the continuous ECG fluctuation data into several time series intervals with similar characteristics, so as to analyze the dynamic characteristics of the fluctuations in each interval. The divided fluctuation intervals can reflect the changing trends and characteristics of the ECG signal in different time periods, which is convenient for subsequent detailed analysis of the ECG fluctuations. Through this interval division, the ability to observe the details of the ECG signal can be improved, which helps to discover the abnormal patterns in the ECG fluctuations, thereby providing basic data support for subsequent abnormal identification. Based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system, the ECG fluctuation morphology data is calculated for the average difference of the fluctuation rising rate in different fluctuation intervals. The calculation is intended to evaluate the change in the rate of fluctuation rise in different fluctuation intervals, so as to identify significant change points. By calculating the average difference in the rising rate of each interval, researchers can obtain quantitative indicators of fluctuation changes, reflecting the dynamic characteristics of cardiac electrical activity. Such analysis not only helps to understand the behavior of ECG fluctuations, but also provides important data basis for subsequent abnormal fluctuation identification, which helps to make a more in-depth assessment of the health status of ECG signals. Capture dynamic changes of fluctuation rise rate mean difference data at multiple scales. Multiscale analysis is a powerful tool that can reveal the characteristics and dynamic behavior of signals at different time scales. By performing multiscale analysis on the rise rate mean difference data, the fluctuation patterns at different time scales in the signal can be found and potential abnormal changes can be captured. This analysis can help researchers identify short-term and long-term trends in ECG fluctuations, provide a more comprehensive perspective for the identification of abnormal ECG fluctuations, and thus provide more accurate data support for heart health monitoring and intervention. Based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system, the regional decline rate ratio of ECG fluctuation morphology data in different fluctuation intervals is calculated. This calculation aims to analyze the rate of fluctuation decline in different intervals and compare their changes. The regional decline rate ratio can reveal the sharp decline of ECG fluctuations in a specific time period, thereby helping to identify potential heart abnormalities. Through quantitative analysis of the decline rate ratio, medical professionals can better understand the electrical activity of the heart, identify pathological changes, and provide a reliable basis for the detection of abnormal ECG fluctuations. According to the ECG fluctuation regional decline rate ratio, the ECG fluctuation morphology data is identified for the decline acceleration area in different fluctuation intervals. By identifying the accelerated descent area of ​​ECG fluctuations within a specific period of time, timely warnings of abnormal cardiac conditions can be provided. This analysis is critical for identifying acute cardiac events or other potential health risks. By accurately identifying the accelerated descent area, medical professionals can conduct targeted monitoring and intervention to ensure the safety and health of patients, while providing important data support for clinical decision-making. Abnormal ECG fluctuations are identified based on multi-scale dynamic change data of the rise rate, the ECG fluctuation area descent rate ratio, and ECG descent acceleration area data.This step integrates the various information obtained in the previous steps, and uses a comprehensive analysis method to more accurately identify abnormal ECG fluctuation signals. This multi-dimensional data analysis method not only improves the accuracy of abnormal identification, but also better captures the potential patterns and trends in complex signals. Through accurate abnormal ECG fluctuation identification.

[0030] Preferably, step S23 includes the following steps:

[0031] Step S231: performing abnormal ECG similarity analysis on the abnormal ECG fluctuation data to obtain abnormal ECG similarity data;

[0032] Step S232: calculating the peak curvature of the abnormal ECG fluctuation data according to the abnormal ECG similarity data to obtain abnormal fluctuation peak curvature data;

[0033] Step S233: performing time series cross-correlation analysis on the abnormal fluctuation peak curvature data to obtain peak curvature time series cross-correlation data;

[0034] Step S234: calculating the peak spacing of the abnormal fluctuation peak curvature data based on the peak curvature time series cross-correlation data to obtain abnormal fluctuation peak spacing data;

[0035] Step S235: performing abnormal clustering analysis based on the peak curvature time series cross-correlation data and the abnormal fluctuation peak spacing data to obtain ECG abnormal clustering data.

[0036] The present invention can identify abnormal patterns in ECG signals by performing abnormal ECG similarity analysis on abnormal ECG fluctuation data. The core of this process is to extract similarity features by comparing ECG waveforms in different time periods, thereby distinguishing normal fluctuations from abnormal fluctuations. This analysis can not only provide basic data for subsequent steps, but also quickly locate potential ECG abnormalities and provide support for clinical decision-making. In addition, the acquisition of abnormal ECG similarity data helps to further optimize and adjust the ECG monitoring system and improve the sensitivity and accuracy of monitoring. By calculating the peak curvature of abnormal ECG similarity data, the characteristics of the peak in the ECG signal are extracted. This calculation involves analyzing the geometric characteristics of the waveform, and the curvature of the peak can reflect the changing trend and amplitude of the ECG signal. Accurate peak curvature data helps to identify important waveform features in the ECG, such as the morphological changes of the QRS complex. By analyzing the peak curvature, a quantitative basis can be provided for the abnormal state of the ECG. Performing a time series change cross-correlation analysis on the abnormal fluctuation peak curvature data helps to reveal the changing law of the peak curvature in the time dimension. Through this analysis, the dynamic characteristics of the ECG signal can be understood, especially the interaction between the generation of abnormal fluctuations and other waveforms. This step can not only find the correlation between the peak curvatures, but also reveal the potential factors that cause ECG abnormalities. The peak spacing calculation is based on the cross-correlation data of the peak curvature time series, which can quantify the peak interval in the abnormal fluctuations. This calculation helps to evaluate the frequency characteristics and fluctuation patterns of the ECG signal. With accurate peak spacing data, we can further understand the electrophysiological activity of the heart, especially in the case of arrhythmia or other cardiac pathology, the change in peak spacing can reflect subtle changes in the health of the heart. Abnormal clustering analysis based on the cross-correlation data of the peak curvature time series and the abnormal fluctuation peak spacing data can comprehensively evaluate the severity and frequency of ECG abnormalities. This analysis provides a deep understanding of abnormal ECG phenomena by integrating multi-dimensional data, and can effectively identify the clustering pattern of abnormal signals.

[0037] Preferably, step S3 comprises the following steps:

[0038] Step S31: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal ECG evolution trend data;

[0039] Step S32: performing trend dispersion analysis on the abnormal ECG evolution trend data to obtain evolution trend dispersion data;

[0040] Step S33: performing sample feature sampling on the abnormal ECG evolution trend data according to the evolution trend discreteness data to obtain abnormal evolution trend sample data.

[0041] The present invention first uses abnormal cluster distribution density data to conduct an in-depth analysis of abnormal ECG fluctuation data to identify the existing abnormal patterns and change trends. Through statistical and machine learning methods, a prediction model can be established to help identify future abnormal evolution trends. This prediction can identify potential health risks in advance and provide an important basis for medical intervention. For example, using the time series analysis method, the fluctuation characteristics in the ECG signal can be effectively captured, and then the evolution trend of ECG abnormalities can be predicted. After obtaining the ECG abnormal evolution trend data, the next step is to perform trend dispersion analysis on these data. The core of this step is to evaluate the degree of dispersion of the evolution trends of different ECG abnormalities, which helps to identify which trends have a large change range or uncertainty. By calculating statistical indicators such as the standard deviation or coefficient of variation of the trend, the stability and reliability of the ECG fluctuation data in the evolution process can be quantitatively reflected. Sample feature sampling is performed on the ECG abnormal evolution trend data according to the evolution trend dispersion data. This step aims to extract representative sample data from the existing ECG abnormal evolution trend to facilitate subsequent analysis and research. Through a reasonable sampling strategy, we can ensure that the selected samples are highly representative and effective, thus providing a reliable data basis for further machine learning or pattern recognition.

[0042] Preferably, step S31 includes the following steps:

[0043] Step S311: performing time series disturbance evolution analysis on the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal cluster disturbance evolution data;

[0044] Step S312: performing nonlinear evolution fitting on the abnormal ECG fluctuation data based on the abnormal clustered disturbance evolution data to obtain abnormal ECG fluctuation evolution fitting data;

[0045] Step S313: performing multi-dimensional disturbance accumulation on the abnormal ECG fluctuation evolution fitting data to obtain abnormal ECG fluctuation multi-dimensional disturbance data;

[0046] Step S314: performing common feature recognition on the abnormal ECG fluctuation multi-dimensional disturbance data according to the abnormal ECG fluctuation evolution fitting data to obtain the abnormal ECG fluctuation common feature data;

[0047] Step S315: predicting the abnormal evolution trend according to the abnormal ECG fluctuation evolution fitting data, the abnormal ECG fluctuation multi-dimensional disturbance data and the abnormal ECG fluctuation common feature data, and obtaining the abnormal ECG fluctuation evolution trend data.

[0048] The present invention performs time series perturbation evolution analysis on abnormal ECG fluctuation data based on abnormal cluster distribution density data. The purpose of this analysis is to identify and capture the perturbation characteristics of ECG signals in the time dimension, thereby revealing the dynamic evolution process of ECG fluctuations. By performing time series analysis on the data, the starting point, development process and possible end point of the abnormal fluctuation can be determined. This process helps to better understand the evolution mechanism of abnormal fluctuations, and can provide a preliminary prediction of future ECG activity trends, enhancing the sensitivity and early warning capabilities of the monitoring system. Based on the abnormal cluster perturbation evolution data obtained from step S311, nonlinear evolution fitting is performed on the abnormal ECG fluctuation data. This process aims to capture the complex patterns and nonlinear characteristics of abnormal ECG fluctuations by establishing a mathematical model, thereby achieving accurate description and simulation of the data. Nonlinear fitting can better adapt to the volatility and irregularity in ECG signals, making the obtained fitting data more real and reliable. This analysis result will provide a solid foundation for subsequent trend prediction and anomaly detection. Multidimensional perturbation accumulation is performed on the abnormal ECG fluctuation evolution fitting data, with the aim of comprehensively evaluating the impact of different perturbation factors on ECG fluctuations. By fusing data from different dimensions, we can more comprehensively understand the complexity of ECG signals and identify the main factors affecting ECG fluctuations and their interactions. This process helps to improve the accuracy of data analysis and ensure that the changes in ECG signals can be accurately captured and understood under the influence of multiple factors. At the same time, the accumulation of multidimensional disturbances also lays a data foundation for subsequent abnormal feature identification and trend prediction, and improves the analysis ability of the monitoring system. Common features of abnormal ECG fluctuation multidimensional disturbance data are identified based on abnormal ECG fluctuation evolution fitting data. This process aims to extract common features from complex ECG fluctuation data to identify the similarities between different abnormal ECG signals. The identification of such common features not only helps to understand the basic characteristics and patterns of ECG abnormalities. The abnormal evolution trend is predicted based on the abnormal ECG fluctuation evolution fitting data, abnormal ECG fluctuation multidimensional disturbance data and common feature data obtained in the previous steps. The core of this stage is to use the multidimensional data model to predict the future trend of ECG signals and then determine the development direction of ECG abnormalities.

[0049] Preferably, step S314 includes the following steps:

[0050] Performing hierarchical time series decoupling on the abnormal ECG fluctuation evolution fitting data to obtain fluctuation evolution fitting hierarchical decoupling data;

[0051] According to the fluctuation evolution fitting hierarchical decoupling data, the correlation factor analysis of abnormal ECG fluctuation multidimensional disturbance data was carried out to obtain the evolution fitting-multidimensional disturbance correlation factor;

[0052] Perform nonlinear regression analysis on the evolution fitting-multidimensional disturbance correlation factor to obtain the evolution-disturbance correlation regression data;

[0053] Common feature identification is performed based on evolution-disturbance correlation regression data to obtain common feature data of abnormal ECG fluctuations.

[0054] The present invention performs hierarchical temporal decoupling on abnormal ECG fluctuation evolution fitting data, with the purpose of decomposing complex ECG fluctuation signals into different levels and time series, so as to better understand its internal change mechanism. Through temporal decoupling, components of different frequencies and amplitudes in the signal can be separated, thereby highlighting the characteristics of each level. This method not only helps to identify potential patterns and trends in ECG fluctuations, but also effectively reduces noise interference in data and improves the accuracy of subsequent analysis. Through hierarchical processing, the contribution of each level to ECG abnormalities can be more clearly seen, thus laying the foundation for the identification of abnormal features. After obtaining the hierarchical decoupling data of fluctuation evolution fitting, the next step is to perform correlation factor analysis on the multidimensional perturbation data of abnormal ECG fluctuations. This process aims to explore the relationship between different fluctuation evolution levels and multidimensional perturbations, and identify the key factors affecting ECG fluctuations. By using statistical analysis techniques, the influence of the evolution of each level on the multidimensional perturbation of ECG signals can be quantified, thereby obtaining the correlation factor between evolution fitting and multidimensional perturbations. This analysis can not only reveal the complex relationship behind ECG fluctuations, but also help to find the root cause of abnormal fluctuations, providing an important reference for subsequent intervention and treatment. After obtaining the evolutionary fitting-multidimensional perturbation correlation factor, nonlinear regression analysis is performed. The main purpose of this step is to establish a mathematical model to quantify the relationship between the correlation factor and the ECG fluctuation and find the best fitting curve. Through nonlinear regression, the complexity of ECG signal changes can be captured more accurately, ensuring that the model not only adapts to linear relationships but also handles the influence of nonlinear features. The obtained evolutionary-perturbation correlation regression data will provide a solid data foundation for further analysis of ECG signals, making the construction of prediction models more robust and reliable. Common feature identification is performed based on evolutionary-perturbation correlation regression data. This process aims to extract common features from complex ECG fluctuation data to reveal the similarities and common patterns between different abnormal ECG signals. By using data mining and pattern recognition techniques, common features that have a significant impact on ECG abnormalities can be identified. This identification not only helps to improve the accuracy of abnormality detection, but also provides valuable basis for subsequent clinical decision-making, making ECG monitoring systems more effective in dealing with complex pathological conditions.

[0055] Preferably, the present invention further provides a system for generating electrocardiogram data, which is used to execute the method for generating electrocardiogram data as described above, and the system for generating electrocardiogram data comprises:

[0056] The ECG waveform analysis module is used to collect the user's ECG signal through the electrode sheet to obtain the user's ECG signal; perform a wave shape analysis on the user's ECG signal to obtain ECG wave shape data;

[0057] The abnormal cluster distribution module is used to identify abnormal ECG fluctuations on ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; and to calculate the cluster distribution density of abnormal ECG fluctuation data to obtain abnormal cluster distribution density data;

[0058] The abnormal evolution trend module is used to predict the abnormal evolution trend of abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal ECG evolution trend data; and to perform sample feature sampling on the abnormal ECG evolution trend data to obtain abnormal evolution trend sample data;

[0059] The iterative learning module is used to iterate the learning of abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; based on the abnormal evolution trend learning data, the ECG fluctuation morphology data is annotated with abnormal ECG, and the data is fed back to the terminal to generate ECG data.

[0060] Preferably, a computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for generating electrocardiogram data as described in any one of 1 to 8 above.

[0061] The beneficial effect of the present invention is that the user's ECG signal is collected through the electrode sheet, which is the basis of electrocardiogram (ECG) monitoring. The electrode sheet can effectively capture the slight changes in the heart's electrical activity and convert it into an electrical signal for analysis. Subsequently, by performing a wave morphology analysis on the collected ECG signal, key features in the ECG, such as heartbeat intervals, peaks and troughs, can be extracted. This process can not only reveal the health status of the user's heart, but also provide a data basis for subsequent abnormal identification. By analyzing the ECG wave morphology data, the characteristics of normal ECG signals can be identified, and standards can be set for the detection of abnormal signals, thereby improving the accuracy and effectiveness of subsequent analysis. Abnormal ECG fluctuations are identified on the ECG wave morphology data obtained by step S1. This process uses advanced algorithms, such as machine learning and statistical analysis, to detect abnormal patterns in ECG waveforms. These abnormal fluctuations are indicators of potential health problems such as arrhythmias and myocardial ischemia. Subsequently, the cluster distribution density calculation of the identified abnormal ECG fluctuation data can help further analyze the distribution and density of these abnormal fluctuations. Through this calculation, the frequency of occurrence, concentrated areas and clinical significance of abnormal fluctuations can be identified. This not only helps to improve the detection rate of abnormal fluctuations, but also provides an important basis for clinical decision-making. According to the abnormal cluster distribution density data, the abnormal evolution trend of abnormal ECG fluctuation data is predicted. This process uses time series analysis and prediction models to identify the future development trend of abnormal fluctuations. A better understanding of the evolution of the patient's ECG state can also provide a basis for early intervention. Subsequently, the predicted ECG abnormal evolution trend data is sampled to extract representative feature samples. By extracting sample features, the accuracy and effectiveness of subsequent analysis can be ensured, laying the foundation for in-depth research on abnormal evolution trends. The realization of this process enhances the dynamic monitoring ability of heart health status, provides a scientific basis for early identification of potential cardiovascular diseases, and helps medical personnel make decisions more scientifically and efficiently. The random forest algorithm is used to learn and iterate the abnormal evolution trend sample data. With its powerful classification and regression capabilities, this algorithm shows excellent results when processing complex ECG fluctuation data. Through continuous learning and optimization, random forest can improve the recognition accuracy of abnormal ECG fluctuations and establish a more accurate abnormal evolution trend model. Based on these learning data, the system will perform abnormal ECG labeling on ECG wave morphology data. This labeling result will be fed back to the terminal to generate ECG data, thereby providing an intuitive and easy-to-understand clinical data display. This technical solution effectively integrates data collection, analysis and feedback, ensuring the timeliness and accuracy of information transmission, which not only improves the intelligent level of ECG monitoring, but also promotes the early detection and intervention of cardiovascular diseases.Therefore, the present invention is an improvement on a traditional method for generating electrocardiogram data, which solves the problems of inaccurate recognition of abnormal electrocardiogram signals and inability to accurately judge the evolution trend of electrocardiogram signals in the traditional method for generating electrocardiogram data, improves the accuracy of identifying abnormal electrocardiogram signals and improves the ability to accurately judge the evolution trend of electrocardiogram signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A schematic diagram of the steps of a method for generating electrocardiogram data;

[0063] Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG.

[0064] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0065] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0066] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0067] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0068] To achieve this, please refer to Figure 1 to Figure 2 , a method for generating electrocardiogram data, the method comprising the following steps:

[0069] Step S1: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal; performing a fluctuation morphology analysis on the user's ECG signal to obtain ECG fluctuation morphology data;

[0070] Step S2: performing abnormal ECG fluctuation recognition on the ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; performing cluster distribution density calculation on the abnormal ECG fluctuation data to obtain abnormal cluster distribution density data;

[0071] Step S3: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain the abnormal ECG evolution trend data; sampling the sample features of the abnormal ECG evolution trend data to obtain the abnormal evolution trend sample data;

[0072] Step S4: Perform learning iteration on the abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; perform abnormal ECG labeling processing on the ECG fluctuation morphology data based on the abnormal evolution trend learning data, and feed it back to the terminal to execute the generation of ECG data.

[0073] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for generating electrocardiogram data of the present invention. In this example, the method for generating electrocardiogram data includes the following steps:

[0074] Step S1: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal; performing a fluctuation morphology analysis on the user's ECG signal to obtain ECG fluctuation morphology data;

[0075] In an embodiment of the present invention, the user's ECG signal is collected by placing multiple electrodes at specific locations on the user's body surface. The layout of the electrodes strictly follows international standards, such as a 12-lead ECG system. The electrodes are connected to a high-precision ECG signal acquisition device through wires. The device can capture the voltage fluctuations of each lead in real time and generate a set of ECG signal data in which the voltage changes over time. During the ECG signal acquisition process, the acquisition device performs preliminary hardware filtering on noise and artifacts, such as removing power supply interference and myoelectric interference to ensure signal clarity. Then, the acquired original ECG signal is subjected to a fluctuation morphology analysis. The analysis is mainly achieved through mathematical tools such as Fourier transform and wavelet transform, which can separate the frequency components in the signal, identify key waveform features such as P waves, QRS complexes, and T waves, and ultimately generate detailed ECG fluctuation morphology data. This data contains an accurate description of the amplitude, duration, and waveform of each ECG cycle.

[0076] Step S2: performing abnormal ECG fluctuation recognition on the ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; performing cluster distribution density calculation on the abnormal ECG fluctuation data to obtain abnormal cluster distribution density data;

[0077] In an embodiment of the present invention, after performing ECG fluctuation morphology analysis, the ECG fluctuation morphology data is then subjected to abnormal identification. This process sets normal range parameters for different waveforms, combines statistical methods and biomedical knowledge, and compares the deviation between the actual values ​​of the P wave, QRS complex, and T wave in each ECG cycle and the normal range values. Abnormal ECG fluctuation identification quickly marks data points that do not conform to the normal fluctuation morphology by applying statistically based anomaly detection algorithms, such as the z-score anomaly detection method. These marked data are abnormal ECG fluctuation data. Next, the cluster distribution density of these abnormal ECG fluctuation data is calculated, and the distribution of abnormal ECG fluctuations is analyzed using the kernel density estimation method (KDE). Kernel density estimation can evaluate the density distribution of data points by assigning a Gaussian kernel function to each abnormal data point, thereby obtaining abnormal cluster distribution density data. These data reflect the frequency and distribution of abnormal ECG events.

[0078] Step S3: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain the abnormal ECG evolution trend data; sampling the sample features of the abnormal ECG evolution trend data to obtain the abnormal evolution trend sample data;

[0079] In an embodiment of the present invention, after obtaining the abnormal cluster distribution density data, the abnormal evolution trend of the abnormal ECG fluctuation data is predicted based on this data. This step analyzes the evolution trend of the abnormal data over time by applying a time series analysis method, such as the ARIMA (AutoRegressive Integrated Moving Average) model. The time series prediction tool performs regression analysis on dense abnormal data points to identify potential periodic and trend changes, thereby generating ECG abnormal evolution trend data. Subsequently, sample feature sampling is performed on the ECG abnormal evolution trend data, and the abnormal evolution trend features in different time intervals are extracted using statistical sampling techniques. The dimensions are reduced through principal component analysis (PCA) to extract the most representative abnormal sample feature data. These abnormal evolution trend sample data are used for further analysis and modeling.

[0080] Step S4: Perform learning iteration on the abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; perform abnormal ECG labeling processing on the ECG fluctuation morphology data based on the abnormal evolution trend learning data, and feed it back to the terminal to execute the generation of ECG data.

[0081] In an embodiment of the present invention, a random forest algorithm is applied to perform learning iterations based on the generated abnormal evolution trend sample data. The random forest algorithm is an integrated learning method that improves the accuracy and robustness of predictions by constructing multiple decision trees. In this process, the sample data is randomly assigned to each decision tree, and each tree analyzes the abnormal ECG features in the data and makes independent predictions, and finally determines the optimal prediction result through a voting mechanism. With multiple iterations, the algorithm will gradually identify the potential rules in the abnormal ECG data and generate abnormal evolution trend learning data. Finally, based on this learning data, each abnormal ECG fluctuation in the ECG fluctuation morphology data is labeled. This process is implemented through an automated labeling tool, which automatically labels abnormal ECG events as different types of abnormalities (such as atrial fibrillation, ventricular fibrillation, etc.) in combination with the labels in the learning data, and finally feeds the labeled ECG data back to the terminal device to generate complete ECG data with labels for further medical diagnosis.

[0082] Preferably, step S1 comprises the following steps:

[0083] Step S11: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal;

[0084] Step S12: performing signal denoising processing on the user's ECG signal to obtain a user's ECG denoised signal;

[0085] Step S13: performing baseline drift removal processing on the user's ECG signal according to the user's ECG denoising signal to obtain the user's ECG baseline drift removal data;

[0086] Step S14: performing fluctuation shape analysis on the user's ECG baseline drift-free data to obtain ECG fluctuation shape data.

[0087] In an embodiment of the present invention, the user's ECG signal is collected by fixing multiple electrode sheets on the user's skin surface according to the standard ECG lead position. The electrode sheet material selects highly conductive silver chloride (AgCl) to ensure the stability of signal transmission. The electrode sheet is connected to the ECG signal acquisition device through the lead wire, and the device can detect the weak electrical signals generated by the user's cardiac electrical activity. The acquisition process uses a high-precision analog-to-digital converter (ADC) to convert the analog ECG signal into a digital signal, and the sampling frequency is usually set at 500Hz or higher to ensure that the subtle changes in each ECG cycle are captured. During the acquisition process, the environmental electromagnetic interference is shielded to avoid the interference of 50Hz power frequency noise. The final result is the unprocessed user ECG signal data, which contains the original cardiac electrical activity information. The user ECG signal collected in step S11 is subjected to signal denoising. Since the ECG signal is easily affected by a variety of external noises, such as electromagnetic interference, myoelectric noise and motion artifacts, a discrete wavelet transform (DWT) is required for denoising. The denoising step first decomposes the ECG signal into multiple scales and decomposes it into multiple different frequency components by selecting a suitable wavelet basis function (such as Daubechies wavelet). Then, the soft threshold denoising method is applied in the high-frequency subband to suppress high-frequency noise interference. The details of the denoised signal are retained, while the noise is effectively removed to obtain the user's ECG denoised signal. This process ensures the smoothness and accuracy of the signal, making subsequent analysis more reliable. After the denoising process is completed, the user's ECG denoised signal is subjected to baseline drift removal. Baseline drift is a low-frequency fluctuation phenomenon caused by breathing, skin resistance changes, etc. An adaptive algorithm based on a sliding average filter is used to remove baseline drift. First, the ECG denoised signal is subjected to a sliding average operation over a large time window (such as 200 milliseconds) to extract the low-frequency drift component of the signal. Then, the low-frequency component is subtracted from the original signal to eliminate the influence of baseline drift. This operation ensures the baseline stability of the ECG signal, making important ECG features such as P wave, QRS complex and T wave clearer, and finally obtaining the user ECG baseline drift-free data. The user ECG baseline drift-free data generated in step S13 is subjected to a fluctuation morphology analysis. The fluctuation morphology analysis mainly depends on the time domain and frequency domain characteristics of the signal. The P wave, QRS complex and T wave in the ECG cycle are accurately located by using a peak detection algorithm. In this process, the fast rising segment of the QRS complex is detected by differential operation and sliding window method, and then the position of the P wave and T wave is reversely inferred based on the position of the QRS complex. Next, the amplitude, duration and time interval of each waveform are quantified to analyze the complete fluctuation morphology of the ECG signal. These analysis results form ECG fluctuation morphology data, which includes detailed characteristic parameters of the ECG cycle for further clinical diagnosis and analysis.

[0088] Preferably, step S2 comprises the following steps:

[0089] Step S21: constructing a Cartesian coordinate system for the ECG wave shape data to obtain an ECG wave shape coordinate system;

[0090] Step S22: performing abnormal ECG fluctuation identification on the ECG fluctuation morphology data according to the ECG fluctuation morphology coordinate system to obtain abnormal ECG fluctuation data;

[0091] Step S23: performing abnormal clustering analysis on the abnormal ECG fluctuation data to obtain abnormal ECG clustering data;

[0092] Step S24: Calculate the cluster distribution density of the abnormal ECG fluctuation data according to the abnormal ECG cluster data to obtain abnormal cluster distribution density data.

[0093] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0094] Step S21: constructing a Cartesian coordinate system for the ECG wave shape data to obtain an ECG wave shape coordinate system;

[0095] In an embodiment of the present invention, ECG wave morphology data is used to construct a Cartesian coordinate system. This step geometrically maps the wave morphology of the ECG signal by establishing a coordinate system of the time axis (horizontal axis) and the ECG voltage value (vertical axis) in a two-dimensional space. First, each ECG cycle in the ECG signal is time-calibrated, using precise time scale units (such as milliseconds) to ensure that the timing changes of the signal are accurately represented on the time axis. Then, the voltage amplitude of the signal is used as a parameter of the vertical axis to calibrate the various characteristic points of the ECG wave, such as the voltage values ​​of the P wave, QRS complex, and T wave. The construction of this coordinate system provides a spatial basis for subsequent wave morphology analysis, forming an ECG wave morphology coordinate system, which clearly shows the multi-dimensional spatial characteristics of the ECG signal in time and voltage changes.

[0096] Step S22: performing abnormal ECG fluctuation identification on the ECG fluctuation morphology data according to the ECG fluctuation morphology coordinate system to obtain abnormal ECG fluctuation data;

[0097] In an embodiment of the present invention, abnormal ECG fluctuations are identified for ECG fluctuation morphology data based on the ECG fluctuation morphology coordinate system in step S21. In this process, the threshold method is used to analyze the characteristic points in the ECG signal, and an abnormal judgment threshold is constructed based on the reference range of the normal ECG signal (such as P wave duration, QRS wave group duration, R wave height, etc.). By checking the peak value, duration, amplitude change and other parameters of each band one by one, the bands that exceed the normal range are marked, and then the abnormal ECG fluctuation data are identified. This recognition process not only focuses on the abnormal amplitude of the signal, but also comprehensively considers abnormal characteristics such as time offset and waveform distortion to ensure that abnormal ECG events can be accurately identified.

[0098] Step S23: performing abnormal clustering analysis on the abnormal ECG fluctuation data to obtain abnormal ECG clustering data;

[0099] In an embodiment of the present invention, an abnormal clustering analysis is performed based on the abnormal ECG fluctuation data. The clustering analysis uses a spatial density analysis method to map multiple bands of abnormal ECG fluctuation data to an ECG fluctuation morphological coordinate system, and perform density clustering analysis on their distribution. This step uses a clustering method based on a nearest neighbor algorithm to analyze whether multiple abnormal ECG events occur in a certain period of time or a certain area by calculating the proximity and degree of clustering of each abnormal band. If there are multiple abnormal fluctuations within a certain period of time, and the time intervals and amplitude change trends of these fluctuations are similar, it is considered to be an abnormal clustering phenomenon. Finally, abnormal ECG clustering data is formed to clearly show the spatiotemporal distribution characteristics of abnormal fluctuations.

[0100] Step S24: Calculate the cluster distribution density of the abnormal ECG fluctuation data according to the abnormal ECG cluster data to obtain abnormal cluster distribution density data.

[0101] In an embodiment of the present invention, based on the ECG abnormal clustering data, the cluster distribution density of the abnormal ECG fluctuation data is calculated. The cluster distribution density calculation adopts the kernel density estimation method. First, the abnormal fluctuation data is smoothed by a convolution operation to ensure the continuity of the data and the consistency of the distribution. Then, in the area where the abnormal fluctuations are clustered, the density of each abnormal ECG event is estimated by applying the kernel function. The density estimation result reflects the frequency and concentration of abnormal events in a certain area. By calculating the density gradient of abnormal fluctuation events in each area, it is possible to distinguish between abnormal cluster dense areas and abnormal dispersed areas, forming abnormal cluster distribution density data.

[0102] Preferably, step S22 comprises the following steps:

[0103] Step S221: dividing the ECG fluctuation morphology data into time series fluctuation intervals according to the ECG fluctuation morphology coordinate system to obtain the ECG fluctuation time series fluctuation intervals;

[0104] Step S222: calculating the mean difference of the fluctuation rising rates of the ECG fluctuation morphology data in different fluctuation intervals based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system to obtain the fluctuation rising rate mean difference data;

[0105] Step S223: Capture the multi-scale dynamic changes of the fluctuation rising rate mean difference data to obtain the multi-scale dynamic change data of the rising rate;

[0106] Step S224: calculating the regional decline rate ratios of different fluctuation intervals of the ECG fluctuation morphology data based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system to obtain the ECG fluctuation regional decline rate ratios;

[0107] Step S225: identifying the descending acceleration regions of different fluctuation intervals of the ECG fluctuation shape data according to the ECG fluctuation region descending rate ratio, and obtaining ECG descending acceleration region data;

[0108] Step S226: Identify abnormal ECG fluctuations based on the multi-scale dynamic change data of the rising rate, the ECG fluctuation area falling rate ratio and the ECG falling acceleration area data to obtain abnormal ECG fluctuation data.

[0109] In the embodiment of the present invention, the ECG fluctuation morphology data is first divided into time series fluctuation intervals by the ECG fluctuation morphology coordinate system. The specific operation includes dividing the entire cycle of the ECG signal into multiple time series intervals, each interval corresponding to different fluctuation stages of the ECG signal, such as the rising period of the P wave, the rapid fluctuation period of the QRS complex, and the falling period of the T wave. This division is based on the natural periodicity and timing characteristics of the ECG fluctuation signal, ensuring that each time series interval can accurately reflect the local fluctuation of the signal and generate the ECG fluctuation time series fluctuation interval. The accuracy of the interval division is adjusted by analyzing the signal sampling frequency and time interval, thereby laying the foundation for subsequent rate calculation and abnormal identification on the basis of accurately calibrating the ECG fluctuation changes. Based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system generated in step S221, the fluctuation rising rate of each time series fluctuation interval is calculated by the mean difference. This step calculates the mean and difference of the rate change by performing point-by-point rate calculation on the fluctuation rising segment of each time series interval. The calculation of the rising rate adopts the derivative method, and the instantaneous rising rate is obtained by derivation of the increment and time increment of each data point. The mean difference calculation is to count the instantaneous rate of each interval, calculate the mean of each rate in the interval and compare it with the actual fluctuation rate difference, and finally obtain the mean difference data of the fluctuation rising rate. This data can accurately reflect the dynamic change degree of the fluctuation rising in each interval. The multi-scale dynamic change capture of the mean difference data of the fluctuation rising rate. The multi-scale dynamic change capture uses the scale decomposition method. First, the mean difference data of the rising rate is decomposed into dynamic change characteristics on different time scales through multi-scale analysis tools such as wavelet transform. At each scale, the short-term fluctuation and long-term trend of the fluctuation rising rate are captured respectively, and the local mutation and overall fluctuation trend of the rate change are identified to obtain the multi-scale dynamic change data of the rising rate. Through this multi-scale decomposition, the change pattern of the fluctuation can be observed more carefully, helping to identify the small but significant abnormal rising rate change characteristics. According to the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system, the regional decline rate ratio is calculated for different fluctuation intervals of the ECG fluctuation morphology data. First, the ECG signal of each decline interval is subjected to derivative operation to obtain the decline rate. Then, the decline rates of adjacent bands are compared to calculate the rate ratio. The rate ratio can reflect the difference in the downward trend of different intervals. If the ratio changes significantly, it means that the decline rate has abnormally changed or slowed down. The calculated decline rate ratio of the ECG fluctuation area accurately quantifies the difference in the decline rate in each interval, providing an important numerical basis for subsequent abnormality detection. Based on the decline rate ratio of the ECG fluctuation area calculated in step S224, the decline acceleration area in the ECG fluctuation morphology is identified. This step uses the rate ratio to classify different time series intervals, looking for areas where the rate ratio suddenly increases or exceeds the threshold, and these areas are considered to be decline acceleration areas. The specific operation includes setting the upper and lower limit thresholds of the rate ratio, and marking the fluctuation interval that exceeds the range as the decline acceleration area.The purpose of this operation is to identify the sharp decline phenomenon in the ECG signal, which is usually related to abnormal ECG activity, and finally generate ECG acceleration area data. Abnormal ECG fluctuations are identified based on the multi-scale dynamic change data of the rising rate, the ECG fluctuation area decline rate ratio and the ECG acceleration area data. In this step, multiple data sources are first comprehensively analyzed, and the abnormal area is cross-validated through the multi-dimensional data of rate changes. If a certain interval shows abnormal fluctuations in the rising rate and abnormal acceleration in the falling rate at the same time, it is marked as an abnormal ECG fluctuation area. During the identification process, each abnormal area is confirmed by combining threshold judgment and trend analysis to ensure the accuracy of identification. Finally, the abnormal ECG fluctuation data is output to clearly locate the spatiotemporal distribution and characteristics of abnormal ECG fluctuations.

[0110] Preferably, step S23 includes the following steps:

[0111] Step S231: performing abnormal ECG similarity analysis on the abnormal ECG fluctuation data to obtain abnormal ECG similarity data;

[0112] Step S232: calculating the peak curvature of the abnormal ECG fluctuation data according to the abnormal ECG similarity data to obtain abnormal fluctuation peak curvature data;

[0113] Step S233: performing time series cross-correlation analysis on the abnormal fluctuation peak curvature data to obtain peak curvature time series cross-correlation data;

[0114] Step S234: calculating the peak spacing of the abnormal fluctuation peak curvature data based on the peak curvature time series cross-correlation data to obtain abnormal fluctuation peak spacing data;

[0115] Step S235: performing abnormal clustering analysis based on the peak curvature time series cross-correlation data and the abnormal fluctuation peak spacing data to obtain ECG abnormal clustering data.

[0116] In an embodiment of the present invention, an abnormal ECG similarity analysis is performed based on abnormal ECG fluctuation data. The similarity analysis mainly includes the matching of the relative amplitude, frequency and fluctuation period of the waveform by making a fine-grained comparison of the morphological features of the abnormal fluctuations. First, the abnormal ECG fluctuation data is standardized to ensure that each data feature is compared at the same numerical scale. Then, the similarity between each abnormal fluctuation is calculated using an algorithm based on a distance metric (such as Euclidean distance or cosine similarity). Through multiple iterative calculations and comparisons, an abnormal fluctuation group with high similarity is screened out to obtain abnormal ECG similarity data. Based on the abnormal ECG similarity data obtained in step S231, the peak curvature of the abnormal ECG fluctuation data is calculated. In this process, the calculation of the peak curvature uses a geometric analysis method to solve the curvature value of the peak through the derivative. The specific operation steps include: first, identifying each peak position in the abnormal fluctuation, determining the trough points around it, and calculating the curvature at the peak by the three-point method. The curvature value is calculated by combining the first-order derivative and the second-order derivative at the peak, reflecting the curvature degree and change rate of the peak. The larger the curvature value, the more drastic the fluctuation. Finally, the abnormal fluctuation peak curvature data is output, which can accurately characterize the drastic change characteristics in abnormal fluctuations. The abnormal fluctuation peak curvature data is subjected to time series change cross-correlation analysis. This analysis method is to combine each peak curvature value with its corresponding time axis data to perform a multi-dimensional cross-analysis of the time series change trend. The specific operation steps are as follows: First, arrange each peak curvature value in chronological order, and capture the curvature changes in different time periods by the sliding window method. Then, compare the correlation of the curvature changes in different time periods, and calculate its time series cross-correlation index. This analysis can reveal the dynamic change relationship of the fluctuation curvature in time, identify the abnormal change trend at a specific time point, and generate peak curvature time series cross-correlation data. The abnormal fluctuation peak curvature data is subjected to time series change cross-correlation analysis. This analysis method is to combine each peak curvature value with its corresponding time axis data to perform a multi-dimensional cross-analysis of the time series change trend. The specific operation steps are as follows: First, arrange each peak curvature value in chronological order, and capture the curvature changes in different time periods by the sliding window method. Then, compare the correlation of curvature changes in different time periods and calculate its time series cross-correlation index. This analysis can reveal the dynamic change relationship of fluctuation curvature in time, identify abnormal change trends at specific time points, and generate peak curvature time series cross-correlation data. Calculate the peak spacing of abnormal fluctuation peak curvature data. This operation step evaluates the fluctuation frequency and fluctuation density by calculating the time interval between peaks. First, extract the peak time point in each abnormal fluctuation and calculate the time spacing between adjacent peaks. Next, perform a statistical analysis on the spacing of all peaks to find the intervals with abnormal spacing and identify abnormally dense or sparse peak distributions.Finally, the abnormal fluctuation peak spacing data is output, which can reveal the abnormal distribution of ECG fluctuations on the time axis and mark the area with abnormal fluctuation frequency. Combined with the peak curvature time series cross-correlation data and the abnormal fluctuation peak spacing data, the abnormal ECG fluctuations are subjected to clustering analysis. First, based on the data generated in the first two steps, the distribution characteristics of the fluctuations on the time axis and frequency are analyzed to find clustering characteristics. The fluctuation density in different areas is calculated by statistical methods, and areas with highly dense or sparse peaks are identified. Clustering analysis uses a clustering algorithm to classify abnormal fluctuations with similar characteristics to identify clustering phenomena. Finally, the ECG abnormal clustering data is output, which clearly identifies the concentrated distribution areas of abnormal fluctuations, providing important spatial and temporal feature information for subsequent ECG analysis.

[0117] Preferably, step S3 comprises the following steps:

[0118] Step S31: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal ECG evolution trend data;

[0119] Step S32: performing trend dispersion analysis on the abnormal ECG evolution trend data to obtain evolution trend dispersion data;

[0120] Step S33: performing sample feature sampling on the abnormal ECG evolution trend data according to the evolution trend discreteness data to obtain abnormal evolution trend sample data.

[0121] In an embodiment of the present invention, the abnormal ECG fluctuation data is predicted for abnormal evolution trend according to the abnormal cluster distribution density data. The process depends on the change trend of the time series data. By predicting the change trajectory of the dense area and the sparse area in the abnormal cluster distribution density data, the development direction of the abnormal fluctuation in the future is obtained. The specific implementation steps are: first, the abnormal ECG fluctuation data is split into time series to determine the abnormal aggregation state in each time period. Then, the polynomial fitting method is used to fit the density change of the cluster area in the time series, and the trend of density change is predicted in combination with derivative analysis. By comprehensively calculating the trend curves of different time periods, an abnormal evolution trend curve is generated, and finally the ECG abnormal evolution trend data is output. This data provides a change prediction of abnormal fluctuations in the time dimension for subsequent analysis. Trend dispersion analysis is performed on the ECG abnormal evolution trend data. The process evaluates the stability and change amplitude of the abnormal trend by calculating the dispersion of the fluctuation amplitude and frequency of different trend segments in the evolution trend curve. First, the ECG abnormal evolution trend data is divided into several time periods, and the fluctuation amplitude and frequency of the trend curve are calculated in each time period. Next, the dispersion calculation formula, such as variance or standard deviation, is used to calculate the fluctuation dispersion value in each time period. This dispersion value can reveal the drastic degree of trend change. The larger the fluctuation, the higher the dispersion value, indicating the instability of abnormal ECG fluctuations in this period. Finally, the evolution trend dispersion data is output, which is used to evaluate the fluctuation range and stability of ECG abnormal evolution. According to the evolution trend dispersion data, sample feature sampling is performed on the ECG abnormal evolution trend data. The goal of this process is to extract key feature samples representing abnormal evolution trends by sampling specific ECG fluctuation data from high and low dispersion areas. The specific operation steps are as follows: First, the evolution trend dispersion data in different time periods are classified to divide high and low dispersion areas. Then, representative ECG fluctuation samples are selected from each dispersion area, and the characteristics such as peak position, trough position and fluctuation amplitude are considered in the sampling process. Then, by comparing the characteristic values ​​of each sample, the main trend characteristics of abnormal ECG evolution are summarized. Finally, abnormal evolution trend sample data is generated, which is used for further feature analysis and abnormal classification.

[0122] Preferably, step S31 includes the following steps:

[0123] Step S311: performing time series disturbance evolution analysis on the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal cluster disturbance evolution data;

[0124] Step S312: performing nonlinear evolution fitting on the abnormal ECG fluctuation data based on the abnormal clustered disturbance evolution data to obtain abnormal ECG fluctuation evolution fitting data;

[0125] Step S313: performing multi-dimensional disturbance accumulation on the abnormal ECG fluctuation evolution fitting data to obtain abnormal ECG fluctuation multi-dimensional disturbance data;

[0126] Step S314: performing common feature recognition on the abnormal ECG fluctuation multi-dimensional disturbance data according to the abnormal ECG fluctuation evolution fitting data to obtain the abnormal ECG fluctuation common feature data;

[0127] Step S315: predicting the abnormal evolution trend based on the abnormal ECG fluctuation evolution fitting data, the abnormal ECG fluctuation multi-dimensional disturbance data and the abnormal ECG fluctuation common feature data to obtain the abnormal ECG evolution trend data.

[0128] In an embodiment of the present invention, the abnormal ECG fluctuation data is subjected to a time series perturbation evolution analysis based on the abnormal cluster distribution density data. First, the time series in the abnormal cluster distribution density data needs to be segmented, the ECG fluctuations are divided into multiple subsequences according to the time axis, and high-density cluster areas and low-density areas are identified in each subsequence. By performing perturbation amplitude and frequency analysis on the time series change trends of these areas, the perturbation evolution model based on the differential equation is used to fit and analyze the perturbation changes in each subsequence, and finally the abnormal cluster disturbance evolution data is obtained. The data contains the time series change characteristics of the abnormal ECG fluctuations, which is used for the subsequent nonlinear evolution fitting step. Based on the abnormal cluster disturbance evolution data, the abnormal ECG fluctuation data is subjected to nonlinear evolution fitting. First, a nonlinear function (such as a Logistic function or an exponential function) is selected to fit the time series curve of the abnormal ECG fluctuation data to describe the nonlinear growth, saturation or decay trend in the ECG fluctuation evolution. In the fitting process, the disturbance feature points (such as peaks, valleys, etc.) in the abnormal cluster disturbance evolution data are used as key nodes, and the nonlinear function parameters are tuned by optimization algorithms such as the gradient descent method, so that the fitting curve accurately reflects the evolution trend of the ECG fluctuation. Finally, the abnormal ECG fluctuation evolution fitting data is generated as the basis for subsequent multidimensional disturbance analysis. Multidimensional disturbance accumulation is performed on the abnormal ECG fluctuation evolution fitting data. This step aims to identify the cumulative effect of disturbances in various dimensions (such as time, amplitude, and frequency) in abnormal ECG fluctuations. First, the abnormal ECG fluctuation evolution fitting data is disassembled according to time periods, and multidimensional features are extracted for the fluctuations in each time period, including the frequency change of the fluctuation, the amplitude fluctuation, and its relationship with time. Through mathematical methods such as accumulation and integration, the fluctuation features of each dimension are accumulated to form a multidimensional disturbance matrix. This matrix reflects the cumulative effect of disturbances in each dimension of ECG fluctuations, and the output is the multidimensional disturbance data of abnormal ECG fluctuations. According to the abnormal ECG fluctuation evolution fitting data, the common features of the abnormal ECG fluctuation multidimensional disturbance data are identified. By clustering the multidimensional perturbation data, using methods such as K-means clustering or hierarchical clustering, ECG fluctuation segments with similar perturbation characteristics are classified to identify common features. The identification of common features is based on the similarity analysis of each dimension of the multidimensional perturbation data, and the focus is on identifying the regularity and repetitiveness between different fluctuation segments. Finally, by analyzing the correlation between the feature points in the abnormal ECG fluctuation evolution fitting data and these common perturbation features, the common feature data of abnormal ECG fluctuations are output to provide a basis for further prediction of abnormal evolution trends. Based on the abnormal ECG fluctuation evolution fitting data, the abnormal ECG fluctuation multidimensional perturbation data, and the abnormal ECG fluctuation common feature data, the abnormal evolution trend is predicted. First, using the trend information in the abnormal ECG fluctuation evolution fitting data, combined with the perturbation accumulation effect in the multidimensional perturbation data, an evolution model based on time series prediction is established.The model predicts the abnormal evolution path in the future time period by combining the nonlinear change trend and the repeatability of common characteristics in abnormal ECG fluctuations. Finally, the historical data of ECG fluctuations are used to fit the trend of future moments to output the final ECG abnormal evolution trend data for further health status analysis and early warning.

[0129] Preferably, step S314 includes the following steps:

[0130] Performing hierarchical time series decoupling on the abnormal ECG fluctuation evolution fitting data to obtain fluctuation evolution fitting hierarchical decoupling data;

[0131] According to the fluctuation evolution fitting hierarchical decoupling data, the correlation factor analysis of abnormal ECG fluctuation multidimensional disturbance data was carried out to obtain the evolution fitting-multidimensional disturbance correlation factor;

[0132] Perform nonlinear regression analysis on the evolution fitting-multidimensional disturbance correlation factor to obtain the evolution-disturbance correlation regression data;

[0133] Common feature identification is performed based on evolution-disturbance correlation regression data to obtain common feature data of abnormal ECG fluctuations.

[0134] In an embodiment of the present invention, the abnormal ECG fluctuation evolution fitting data is subjected to layered processing by a time series decoupling algorithm. Specifically, the ECG fluctuation time series signal is layered according to the time interval, waveform characteristics and the strength of the abnormal fluctuation. The time series data of each layer show different fluctuation characteristics. For example, a certain layer focuses on reflecting the long-period anomaly of low-frequency fluctuations, while another layer reflects the short-period anomaly of high-frequency fluctuations. The data of each layer after decoupling processing exists independently, maintaining its distribution characteristics in time and frequency. Finally, the outputted fluctuation evolution fitting layered decoupling data can accurately reflect the ECG fluctuation characteristics of different levels, providing a basis for the multidimensional perturbation analysis in the subsequent steps. The fluctuation evolution fitting layered decoupling data is compared and analyzed with the abnormal ECG fluctuation multidimensional perturbation data to identify the correlation factor between the two. The specific operation is to establish a multidimensional perturbation feature vector, map it with the feature vector of the layered decoupling data, and extract the dimensional features with strong correlation between the two by calculating the covariance matrix of the two. Combined with multivariate statistical methods such as principal component analysis (PCA), factor dimension reduction and analysis are performed on the fitting decoupling data and multidimensional perturbation data at different levels, and finally the evolutionary fitting-multidimensional perturbation correlation factors that can describe the correlation between the two are obtained. These factors play an important role in characterizing the key interactions between abnormal fluctuations and perturbation behaviors. After the correlation factors are extracted, these factors are regressed and analyzed using a nonlinear regression model. Specifically, a suitable nonlinear regression function (such as an exponential, power, or high-order polynomial function) is selected, and the evolutionary fitting-multidimensional perturbation correlation factor is used as an independent variable to regress and fit the nonlinear relationship between the evolutionary trend of ECG fluctuations and the multidimensional perturbation characteristics. The parameters of the regression function are optimized and iterated by the least squares method or gradient descent method until the model converges. This regression analysis aims to reveal the cumulative impact of various correlation factors on ECG abnormal fluctuations during the evolution of ECG fluctuations. The evolutionary-perturbation correlation regression data finally generated contains the precise functional relationship between the ECG fluctuation trend and the multidimensional perturbation, providing support for the identification of common features. Finally, common features are identified based on the evolutionary-perturbation correlation regression data. The specific steps include, first, cross-analyzing the ECG fluctuation data with its corresponding regression data to extract ECG segments with similar fluctuation patterns and disturbance characteristics. Pattern recognition technology (such as feature extraction method based on template matching) is used to identify the common characteristics of frequently occurring abnormal fluctuations. In this process, core parameters such as the peak value, frequency, and amplitude change of the fluctuation are focused on. These common characteristics are further classified into different ECG fluctuation patterns, and each type of common characteristics represents a typical abnormal ECG fluctuation form. The common characteristic data of abnormal ECG fluctuations finally outputted accurately describes the universal characteristics of abnormal ECG fluctuations, providing a reliable data basis for clinical diagnosis and abnormal monitoring.

[0135] Preferably, the present invention further provides a system for generating electrocardiogram data, which is used to execute the method for generating electrocardiogram data as described above, and the system for generating electrocardiogram data comprises:

[0136] The ECG waveform analysis module is used to collect the user's ECG signal through the electrode sheet to obtain the user's ECG signal; perform a wave shape analysis on the user's ECG signal to obtain ECG wave shape data;

[0137] The abnormal cluster distribution module is used to identify abnormal ECG fluctuations on ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; and to calculate the cluster distribution density of abnormal ECG fluctuation data to obtain abnormal cluster distribution density data;

[0138] The abnormal evolution trend module is used to predict the abnormal evolution trend of abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal ECG evolution trend data; and to perform sample feature sampling on the abnormal ECG evolution trend data to obtain abnormal evolution trend sample data;

[0139] The iterative learning module is used to iterate the learning of abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; based on the abnormal evolution trend learning data, the ECG fluctuation morphology data is annotated with abnormal ECG, and the data is fed back to the terminal to generate ECG data.

[0140] Preferably, a computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for generating electrocardiogram data as described in any one of 1 to 8 above.

[0141] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0142] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for generating electrocardiogram data, characterized in that: The following steps are involved: Step S1: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal; performing a fluctuation morphology analysis on the user's ECG signal to obtain ECG fluctuation morphology data; Step S2: performing abnormal ECG fluctuation recognition on the ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; Perform cluster distribution density calculation on abnormal ECG fluctuation data to obtain abnormal cluster distribution density data; wherein step S2 includes: Step S21: constructing a Cartesian coordinate system for the ECG wave shape data to obtain an ECG wave shape coordinate system; Step S22: performing abnormal ECG fluctuation identification on the ECG fluctuation morphology data according to the ECG fluctuation morphology coordinate system to obtain abnormal ECG fluctuation data; Step S23: performing abnormal clustering analysis on the abnormal ECG fluctuation data to obtain abnormal ECG clustering data; wherein step S23 includes: Step S231: performing abnormal ECG similarity analysis on the abnormal ECG fluctuation data to obtain abnormal ECG similarity data; Step S232: calculating the peak curvature of the abnormal ECG fluctuation data according to the abnormal ECG similarity data to obtain abnormal fluctuation peak curvature data; Step S233: performing time series cross-correlation analysis on the abnormal fluctuation peak curvature data to obtain peak curvature time series cross-correlation data; Step S234: calculating the peak spacing of the abnormal fluctuation peak curvature data based on the peak curvature time series cross-correlation data to obtain abnormal fluctuation peak spacing data; Step S235: performing abnormal clustering analysis based on the peak curvature time series cross-correlation data and the abnormal fluctuation peak interval data to obtain ECG abnormal clustering data; Step S24: calculating the cluster distribution density of the abnormal ECG fluctuation data according to the abnormal ECG cluster data to obtain abnormal cluster distribution density data; Step S3: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain the abnormal ECG evolution trend data; sampling the sample features of the abnormal ECG evolution trend data to obtain the abnormal evolution trend sample data; Step S4: Perform learning iteration on the abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; perform abnormal ECG labeling processing on the ECG fluctuation morphology data based on the abnormal evolution trend learning data, and feed it back to the terminal to execute the generation of ECG data.

2. The method for generating electrocardiogram data according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: collecting the user's ECG signal through the electrode sheet to obtain the user's ECG signal; Step S12: performing signal denoising processing on the user's ECG signal to obtain a user's ECG denoised signal; Step S13: performing baseline drift removal processing on the user's ECG signal according to the user's ECG denoising signal to obtain the user's ECG baseline drift removal data; Step S14: performing fluctuation shape analysis on the user's ECG baseline drift-free data to obtain ECG fluctuation shape data.

3. The method for generating electrocardiogram data according to claim 1, characterized in that: Step S22 includes the following steps: Step S221: dividing the ECG fluctuation morphology data into time series fluctuation intervals according to the ECG fluctuation morphology coordinate system to obtain the ECG fluctuation time series fluctuation intervals; Step S222: calculating the mean difference of the fluctuation rising rates of the ECG fluctuation morphology data in different fluctuation intervals based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system to obtain the fluctuation rising rate mean difference data; Step S223: Capture the multi-scale dynamic changes of the fluctuation rising rate mean difference data to obtain the multi-scale dynamic change data of the rising rate; Step S224: calculating the regional decline rate ratios of different fluctuation intervals of the ECG fluctuation morphology data based on the ECG fluctuation time series fluctuation interval and the ECG fluctuation morphology coordinate system to obtain the ECG fluctuation regional decline rate ratios; Step S225: identifying the descending acceleration regions of different fluctuation intervals of the ECG fluctuation shape data according to the ECG fluctuation region descending rate ratio, and obtaining ECG descending acceleration region data; Step S226: Identify abnormal ECG fluctuations based on the multi-scale dynamic change data of the rising rate, the ECG fluctuation area falling rate ratio and the ECG falling acceleration area data to obtain abnormal ECG fluctuation data.

4. The method for generating electrocardiogram data according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: predicting the abnormal evolution trend of the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal ECG evolution trend data; Step S32: performing trend dispersion analysis on the abnormal ECG evolution trend data to obtain evolution trend dispersion data; Step S33: performing sample feature sampling on the abnormal ECG evolution trend data according to the evolution trend discreteness data to obtain abnormal evolution trend sample data.

5. The method for generating electrocardiogram data according to claim 4, characterized in that: Step S31 includes the following steps: Step S311: performing time series disturbance evolution analysis on the abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal cluster disturbance evolution data; Step S312: performing nonlinear evolution fitting on the abnormal ECG fluctuation data based on the abnormal clustered disturbance evolution data to obtain abnormal ECG fluctuation evolution fitting data; Step S313: performing multi-dimensional disturbance accumulation on the abnormal ECG fluctuation evolution fitting data to obtain abnormal ECG fluctuation multi-dimensional disturbance data; Step S314: performing common feature recognition on the abnormal ECG fluctuation multi-dimensional disturbance data according to the abnormal ECG fluctuation evolution fitting data to obtain the abnormal ECG fluctuation common feature data; Step S315: predicting the abnormal evolution trend based on the abnormal ECG fluctuation evolution fitting data, the abnormal ECG fluctuation multi-dimensional disturbance data and the abnormal ECG fluctuation common feature data to obtain the abnormal ECG evolution trend data.

6. The method for generating electrocardiogram data according to claim 5, characterized in that: Step S314 includes the following steps: Performing hierarchical time series decoupling on the abnormal ECG fluctuation evolution fitting data to obtain fluctuation evolution fitting hierarchical decoupling data; According to the fluctuation evolution fitting hierarchical decoupling data, the correlation factor analysis of abnormal ECG fluctuation multidimensional disturbance data was carried out to obtain the evolution fitting-multidimensional disturbance correlation factor; Perform nonlinear regression analysis on the evolution fitting-multidimensional disturbance correlation factor to obtain the evolution-disturbance correlation regression data; Common feature identification is performed based on evolution-disturbance correlation regression data to obtain common feature data of abnormal ECG fluctuations.

7. A system for generating electrocardiogram data, characterized in that: For executing the method for generating electrocardiogram data according to claim 1, the system for generating electrocardiogram data comprises: The ECG waveform analysis module is used to collect the user's ECG signal through the electrode sheet to obtain the user's ECG signal; perform a wave shape analysis on the user's ECG signal to obtain ECG wave shape data; The abnormal cluster distribution module is used to identify abnormal ECG fluctuations on ECG fluctuation morphology data to obtain abnormal ECG fluctuation data; and to calculate the cluster distribution density of abnormal ECG fluctuation data to obtain abnormal cluster distribution density data; The abnormal evolution trend module is used to predict the abnormal evolution trend of abnormal ECG fluctuation data according to the abnormal cluster distribution density data to obtain abnormal ECG evolution trend data; and to perform sample feature sampling on the abnormal ECG evolution trend data to obtain abnormal evolution trend sample data; The iterative learning module is used to iterate the learning of abnormal evolution trend sample data through the random forest algorithm to obtain abnormal evolution trend learning data; based on the abnormal evolution trend learning data, the ECG fluctuation morphology data is annotated with abnormal ECG, and the data is fed back to the terminal to generate ECG data.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for generating electrocardiogram data as claimed in any one of claims 1 to 6.

Images