A dynamic electrocardiogram data processing method and system

Abstract

The present invention relates to the technical field of dynamic electrocardiogram analysis, specifically a dynamic electrocardiogram data processing method and system, comprising the following steps: segmenting the acquired dynamic electrocardiogram data, adjusting the signal strength of each segment, calculating the energy distribution of each segment, and judging its stability, and generating stability-adjusted electrocardiogram data segments. The present invention, by extracting bioinformatics features based on deep learning for electrocardiogram signals, can identify complex patterns and features in electrocardiogram signals in the early detection of abnormal heart rhythms, thereby improving the accuracy and sensitivity of abnormal heart rhythm identification. It provides strong support for the accurate classification of abnormal heart rhythms. The fractal dimension theory is used to conduct an in-depth analysis of dynamic electrocardiogram data, revealing the nonlinear characteristics and self-similarity of electrocardiogram signals. The theoretical depth of electrocardiogram analysis is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic electrocardiogram analysis, and in particular to a dynamic electrocardiogram data processing method and system. Background Art

[0002] Holter monitors and analyzes ECG data in real time. The core of the field of dynamic ECG analysis technology is to extract, process and interpret ECG signals to diagnose heart disease, monitor heart health, and evaluate treatment effects. It can also capture arrhythmias, abnormal heartbeat intervals, and other heart problems that may not be detected in static ECG examinations.

[0003] Holter data processing refers to a series of technical steps to collect, analyze and interpret ECG data. The purpose is to monitor heart activity in real time or quasi-real time, so as to immediately detect arrhythmia, signs of heart disease or evaluate the effectiveness of heart therapy. It allows doctors to obtain more in-depth heart health information, make more accurate diagnoses, and adjust treatment plans in a timely manner. In effect, Holter data processing aims to improve the accuracy and timeliness of heart disease diagnosis and provide patients with personalized medical solutions.

[0004] Although existing dynamic electrocardiogram technology has made significant progress in real-time capture of ECG data, monitoring of heart health, and evaluation of treatment effects. However, traditional methods mainly focus on linear features and periodic patterns, making it difficult to accurately identify early abnormal heart rhythms or subtle nonlinear changes, affecting the accuracy of diagnosis and early warning capabilities. In addition, when faced with large-scale ECG data, existing technologies lack sufficiently sensitive and accurate bioinformatics feature extraction capabilities, resulting in certain limitations in the early diagnosis of heart disease and the formulation of treatment recommendations. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a dynamic electrocardiogram data processing method and system.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: a method for processing dynamic electrocardiogram data, comprising the following steps;

[0007] S1: Segment the acquired dynamic electrocardiogram data, adjust the signal strength of each segment, calculate the energy distribution of each segment, and determine its stability to generate the electrocardiogram data segment after stability adjustment;

[0008] S2: for the ECG data segment after the stability adjustment, set the initial value of the parameter of each segment of the dynamic ECG data, and adjust the parameter according to the characteristics of each segment to reflect the change of the ECG signal and generate a parameter update record;

[0009] S3: using the parameters to update the record, setting features for various abnormal heart rhythm conditions, analyzing the trend of the ECG signal, matching the signal changes with the features, identifying potential abnormal heart rhythms, and generating abnormal heart rhythm analysis records;

[0010] S4: Based on the recognition record of the abnormal heart rhythm, pattern matching is performed on the electrocardiogram data segment, and it is compared with the known abnormal heart rhythm pattern, and the abnormal heart rhythm pattern similar to the electrocardiogram signal is identified by calculating the similarity of each matching result, and the abnormal heart rhythm pattern similarity analysis result is generated;

[0011] S5: using the similarity analysis result of the abnormal heart rhythm pattern, analyzing the abnormal heart rhythm patterns with similarity, constructing a feature vector for each abnormal heart rhythm, identifying the abnormal heart rhythm features in the electrocardiogram signal, and generating an analysis result of the abnormal heart rhythm type;

[0012] S6: Analyze the analysis results of the abnormal heart rhythm type, measure the variability index of the electrocardiogram signal in each segment, analyze the change pattern of the index, reveal the nonlinear characteristics and hidden abnormal patterns in the signal, and generate signal complexity analysis results.

[0013] The present invention has the following improvements: the stability-adjusted electrocardiogram data segment includes the adjusted signal amplitude, signal stability rating and energy distribution characteristics of each segment; the parameter update record includes the adjusted time window size, signal frequency response adjustment value and signal variability adjustment index; the abnormal heart rhythm analysis record includes abnormal heart rhythm prediction probability, abnormal feature strength judgment and abnormal trend indicator; the abnormal heart rhythm pattern similarity analysis result includes abnormal heart rhythm matching score, matching pattern identification and pattern similarity level; the abnormal heart rhythm type analysis result includes abnormal heart rhythm type confirmation, abnormal feature vector matching degree and abnormal category recognition confidence; the signal complexity analysis result includes signal variability index details, segment nonlinear feature overview and potential abnormal pattern signal characteristics.

[0014] The present invention is improved in that the acquired dynamic electrocardiogram data is segmented, the signal strength of each segment is adjusted, the energy distribution of each segment is calculated, and its stability is determined. The specific steps of generating the electrocardiogram data segment after stability adjustment are as follows;

[0015] S101: segmenting the acquired dynamic electrocardiogram data, adjusting the signal strength segment by segment, calculating the energy distribution of each segment, and obtaining an adjusted signal segment;

[0016] S102: Analyze each of the adjusted signal segments, determine the stability of the signal peak value, mean value, and energy distribution, iteratively adjust the signal strength, and output a segment with optimized signal strength and stability;

[0017] S103: Based on the signal strength and stability optimized segments, the stability of each segment of dynamic electrocardiogram data is analyzed by comparing the energy distribution difference before and after adjustment, and the stability-adjusted electrocardiogram data segment is generated.

[0018] The present invention is improved in that, for the ECG data segment after stability adjustment, the initial value of the parameter of each segment of dynamic ECG data is set, and the parameter is adjusted according to the characteristics of each segment to reflect the change of the ECG signal, and the specific steps of generating the parameter update record are as follows;

[0019] S201: for the electrocardiogram data segments after the stability adjustment, set a set of initial parameters for each segment of data, including basic properties of the signal, including amplitude and frequency response, and create an initial parameter setting record;

[0020] S202: Analyze the characteristics of each segment of data according to the initial parameter setting record, including the change amplitude and frequency characteristics of the signal, adjust the parameters, reflect the characteristics of each segment of the signal, and form a parameter modification record;

[0021] S203: According to the parameter modification record, referring to the signal variability, the requirement of the specified frequency response and the time window, the parameters of each data segment are adjusted again to reflect the current changes of the ECG signal, and a parameter update record is obtained.

[0022] The present invention is improved in that the specific steps of using the parameter to update the record, setting features for various abnormal heart rhythm conditions, analyzing the changing trend of the electrocardiogram signal, matching the signal change with the features, identifying potential abnormal heart rhythm, and generating abnormal heart rhythm analysis records are as follows;

[0023] S301: Collect the parameter update records, define a specified signal feature for each abnormal heart rhythm, reflect the typical changes of the electrocardiogram signal through the feature, and obtain an abnormal heart rhythm feature definition table;

[0024] S302: Based on the abnormal heart rhythm feature definition table, analyzing the change trend of the ECG signal in multiple time windows, combining the local sequence alignment algorithm and the recurrent neural network, matching the change with the features in the table, identifying the signal change matching the known abnormal heart rhythm feature, and obtaining abnormal heart rhythm matching information;

[0025] The local sequence alignment algorithm is according to Formula I:

[0026] ;

[0027] The optimal local alignment score between the ECG signal segment and the defined abnormal heart rhythm features is calculated, and the time series characteristics of the ECG signal are analyzed and its long-term dependencies are learned by combining the local sequence alignment results and the learned pattern features to obtain the abnormal heart rhythm matching information.

[0028] in, The sequence alignment score of the reference ECG signal feature is at position and , is the original sequence at position and The maximum alignment score of , without reference to ECG signal features, For location ECG signal paragraph and location Symptoms of abnormal heart rhythm The matching score between are the QRS complex widths , heart rate fluctuations , Waveform variation , signal strength The weight coefficient of The weight coefficient is determined by analyzing the influence of QRS complex wave width on the accuracy of abnormal heart rhythm recognition. The weight coefficient is determined by statistically analyzing the correlation strength between heart rate fluctuations and each type of heart rhythm abnormality. The weight coefficient is determined by evaluating the ability of waveform variability to distinguish between various abnormal heart rhythms. The weight coefficient is determined by comparing the difference in signal strength between normal ECG and abnormal ECG. is the width of the QRS complex, The fluctuation of heart rate, is the waveform variability, is the signal strength, is the gap penalty value referring to the ECG signal characteristics;

[0029] S303: Summarize the contents of the abnormal heart rhythm matching information, analyze the frequency and intensity of each abnormal heart rhythm feature, identify potential abnormal heart rhythm types accordingly, and generate an abnormal heart rhythm analysis record.

[0030] The present invention is improved in that, based on the recognition record of abnormal heart rhythm, pattern matching is performed on the electrocardiogram data segment, and it is compared with the known abnormal heart rhythm pattern, and the abnormal heart rhythm pattern of the same type as the electrocardiogram signal is identified by calculating the similarity of each matching result, and the specific steps of generating the abnormal heart rhythm pattern similarity analysis result are as follows;

[0031] S401: using the abnormal heart rhythm analysis record, arranging the abnormal heart rhythm information identified in the electrocardiogram data segment, including the abnormal type and its signal characteristics, to obtain an abnormal heart rhythm identification information table;

[0032] S402: performing pattern matching on each electrocardiogram data segment according to the abnormal heart rhythm identification information table, comparing the pattern with known abnormal heart rhythm patterns in the database, recording the matching degree of each pattern, and obtaining a pattern matching degree record table;

[0033] S403: Analyze the pattern matching record table, calculate the similarity of each matching result, select the abnormal heart rhythm pattern that matches the electrocardiogram signal, and generate the abnormal heart rhythm pattern similarity analysis result.

[0034] The present invention is improved in that the specific steps of using the similarity analysis result of the abnormal heart rhythm pattern to analyze the abnormal heart rhythm patterns with similarity, constructing the feature vector of each abnormal heart rhythm, and identifying the abnormal heart rhythm features in the electrocardiogram signal to generate the analysis result of the abnormal heart rhythm type are as follows;

[0035] S501: Based on the similarity analysis result of the abnormal heart rhythm pattern, screen abnormal heart rhythm patterns with similarity, classify the pattern features, and obtain an initial abnormal heart rhythm pattern classification result;

[0036] S502: Analyze the characteristic vector of each type of abnormal heart rhythm according to the initial abnormal heart rhythm pattern classification result, including multiple factors such as frequency, amplitude and time characteristics, to obtain a characteristic vector model;

[0037] S503: Using the feature vector model, by comparing the similarity between the electrocardiogram signal and the feature vector of the abnormal heart rhythm, the feature of the abnormal heart rhythm in the electrocardiogram signal is identified, and an analysis result of the abnormal heart rhythm type is generated.

[0038] The present invention is improved in that the analysis results of the abnormal heart rhythm type are analyzed, the variability index of the electrocardiogram signal in each section is measured, the changing rules of the index are analyzed, the nonlinear characteristics and hidden abnormal patterns in the signal are revealed, and the specific steps of generating the signal complexity analysis results are as follows;

[0039] S601: Based on the analysis result of the abnormal heart rhythm type, preprocess the electrocardiogram data, use the box dimension algorithm to analyze the complexity and self-similarity of the electrocardiogram signal, refer to the fractal dimension of the signal, combine the machine learning algorithm, identify and classify the abnormal heart rhythm pattern in the electrocardiogram signal, and obtain the variability index measurement result;

[0040] The box dimension algorithm is according to Formula II:

[0041] ;

[0042] Calculate the fractal dimension of the ECG signal and generate the complexity and self-similarity values ​​of the ECG signal;

[0043] in, is the fractal dimension of the ECG signal, For scale The minimum number of boxes required to cover the ECG signal is: is the side length of the box, is the heart rate, To correct the QT interval, evaluate the length adjustment of the QT interval in the ECG signal, For ST segment deviation, is the P wave duration, are weight coefficients, corresponding to heart rate , Corrected QT interval , ST segment deviation , P wave duration The importance of The weight coefficient is determined by performing regression analysis on historical ECG data to evaluate the contribution of QRS complex width in abnormal heart rhythm detection. The weight coefficients were determined by analyzing the correlation between heart rate fluctuations and known types of abnormal heart rhythms using Pearson correlation coefficient analysis. The weight coefficients were determined by using variance analysis through multiple case studies of waveform variation in ECG signals to analyze their role in distinguishing various abnormal heart rhythm conditions. The weight coefficient is determined by studying the difference in signal strength between normal and abnormal ECGs, applying a decision tree classifier to compare and analyze the results, and selecting the optimal threshold;

[0044] S602: iteratively analyzing the variability index measurement result, analyzing the change rule of the ECG signal variability index, identifying the nonlinear characteristics of the signal, and obtaining the nonlinear characteristic analysis result;

[0045] S603: Based on the nonlinear feature analysis result, the hidden abnormal pattern in the ECG signal is revealed, and the relationship between the pattern and the known abnormal heart rhythm is determined through comparative analysis to generate a signal complexity analysis result.

[0046] A dynamic electrocardiogram data processing system, the system comprising:

[0047] The data segmentation module divides the dynamic electrocardiogram data into several data segments according to the time series or the number of data points. The amount of data included in each segment can be independently analyzed to obtain a data segment set;

[0048] The stability analysis module performs signal strength adjustment on each segment in the data segment set, calculates energy distribution, and makes stability judgment on each segment to generate a stability analysis result;

[0049] The parameter adjustment module sets initial parameters for each data segment according to the stability analysis results, and adjusts the parameters according to the signal characteristics of each segment to reflect the changing characteristics of the ECG signal and obtain a parameter update record;

[0050] The heart rhythm abnormality feature matching module uses the parameter to update the record, analyzes the changes in the ECG signal for the defined heart rhythm abnormality features, matches the changes with the features, identifies the matched heart rhythm abnormality, and constructs the heart rhythm abnormality analysis record;

[0051] The pattern similarity analysis module compares the electrocardiogram data segment with the known abnormal heart rhythm pattern based on the abnormal heart rhythm analysis record, calculates the similarity of the match, screens the abnormal heart rhythm pattern that matches the electrocardiogram signal, and generates a similarity analysis result;

[0052] The feature vector construction module analyzes the abnormal heart rhythm patterns selected in the similarity analysis results, summarizes the feature vectors of each abnormal heart rhythm, including multiple characteristics such as frequency, amplitude and time, and obtains a set of abnormal heart rhythm feature vectors;

[0053] The signal complexity analysis module measures the variability index of the ECG signal in each segment based on the abnormal rhythm feature vector set, analyzes the variation pattern, reveals the nonlinear characteristics and potential abnormal patterns in the signal, and establishes the signal complexity analysis results.

[0054] Compared with the prior art, the advantages and positive effects of the present invention are:

[0055] In the present invention, by extracting bioinformatics features based on deep learning from electrocardiogram signals, complex patterns and features in electrocardiogram signals can be identified in the early detection of abnormal heart rhythms, thereby improving the accuracy and sensitivity of abnormal heart rhythm identification. This provides strong support for the accurate classification of abnormal heart rhythms. The fractal dimension theory is used to conduct an in-depth analysis of dynamic electrocardiogram data, revealing the nonlinear characteristics and self-similarity of electrocardiogram signals. This not only enhances the theoretical depth of electrocardiogram analysis, but also provides another way to reveal and identify hidden abnormal heart rhythm patterns by quantifying the complexity of electrocardiogram signals. This also brings new possibilities for the formulation of early diagnosis and treatment recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A flow chart of a method for processing dynamic electrocardiogram data is proposed for the present invention;

[0057] Figure 2 The present invention proposes a schematic diagram of the detailed flow of step S1 in a method for processing dynamic electrocardiogram data;

[0058] Figure 3The present invention proposes a schematic diagram of the detailed flow of step S2 in a method for processing dynamic electrocardiogram data;

[0059] Figure 4 The present invention proposes a schematic diagram of the detailed flow of step S3 in a method for processing dynamic electrocardiogram data;

[0060] Figure 5 The present invention proposes a schematic diagram of a detailed flow chart of step S4 in a method for processing dynamic electrocardiogram data;

[0061] Figure 6 The present invention proposes a schematic diagram of a detailed flow chart of step S5 in a method for processing dynamic electrocardiogram data;

[0062] Figure 7 The present invention proposes a schematic diagram of the detailed flow of step S6 in a method for processing dynamic electrocardiogram data;

[0063] Figure 8 A module diagram of a dynamic electrocardiogram data processing system is proposed for the present invention. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0065] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating positions or positional relationships, are based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0066] Example: See Figure 1 ,The present invention provides a technical solution: a method for processing dynamic electrocardiogram data, comprising the following steps;

[0067] S1: Segment the acquired dynamic electrocardiogram data, adjust the signal strength of each segment, calculate the energy distribution of each segment, and determine its stability to generate the electrocardiogram data segment after stability adjustment;

[0068] S2: for the ECG data segment after stability adjustment, set the initial value of the parameter of each segment of dynamic ECG data, and adjust the parameter according to the characteristics of each segment to reflect the change of ECG signal and generate parameter update record;

[0069] S3: Using parameter update records, setting features for various heart rhythm abnormalities, analyzing the trend of ECG signals, matching signal changes with features, identifying potential heart rhythm abnormalities, and generating heart rhythm abnormality analysis records;

[0070] S4: Based on the recognition record of abnormal heart rhythm, pattern matching is performed on the electrocardiogram data segment, and it is compared with the known abnormal heart rhythm pattern, and the abnormal heart rhythm pattern similar to the electrocardiogram signal is identified by calculating the similarity of each matching result, and the abnormal heart rhythm pattern similarity analysis result is generated;

[0071] S5: using the similarity analysis results of abnormal heart rhythm patterns, analyzing abnormal heart rhythm patterns with similarity, constructing feature vectors for each abnormal heart rhythm, identifying abnormal heart rhythm features in the electrocardiogram signal, and generating analysis results of abnormal heart rhythm types;

[0072] S6: Analyze the analysis results of the abnormal heart rhythm types, measure the variability index of the ECG signal in each segment, analyze the changing rules of the index, reveal the nonlinear characteristics and hidden abnormal patterns in the signal, and generate signal complexity analysis results.

[0073] The stability-adjusted ECG data segments include the adjusted signal amplitude, signal stability rating, and energy distribution characteristics of each segment. The parameter update records include the adjusted time window size, signal frequency response adjustment value, and signal variability adjustment index. The abnormal rhythm analysis records include the abnormal rhythm prediction probability, abnormal feature strength judgment, and abnormal trend indicator. The abnormal rhythm pattern similarity analysis results include the abnormal rhythm matching score, the matched pattern identification, and the pattern similarity level. The abnormal rhythm type analysis results include the abnormal rhythm type confirmation, the abnormal feature vector matching degree, and the abnormal category recognition confidence. The signal complexity analysis results include the signal variability index details, the segment nonlinear feature overview, and the potential abnormal pattern signal characteristics.

[0074] See also Figure 2 , the acquired dynamic electrocardiogram data is segmented, the signal strength of each segment is adjusted, and the energy distribution of each segment is calculated and its stability is determined. The specific steps of generating the electrocardiogram data segment after stability adjustment are as follows;

[0075] S101: Segment the acquired dynamic electrocardiogram data, adjust the signal strength segment by segment, calculate the energy distribution of each segment, and obtain the specific process of the adjusted signal segment as follows;

[0076] In sub-step S101, based on the acquired dynamic electrocardiogram data, a signal segmentation method is adopted, and array operations are performed using Python's numpy library to cut the continuous electrocardiogram signal data into segments of fixed length. The length of each segment is set to 1000 data points. The signal strength is adjusted segment by segment through the normalization function scale. The normalization function scale processes the signal strength in a normalized manner with a mean of 0 and a standard deviation of 1. At the same time, the energy calculation function is used to calculate the energy distribution of each segment. The function estimates the energy based on the sum of the squares of the signal to obtain the adjusted signal segment.

[0077] S102: Analyze each adjusted signal segment, determine the stability of the signal peak value, mean value and energy distribution, iteratively adjust the signal strength, and output the specific process of the segment with optimized signal strength and stability as follows;

[0078] In sub-step S102, based on the adjusted signal segment, a stability evaluation method is adopted, and the signal module in the SciPy library is used to calculate the peak value, mean value and energy distribution of the signal. The ratio of the standard deviation and mean value of the segment signal is calculated as the stability rating indicator, and the signal strength is iteratively adjusted. The iterative adjustment adopts the successive approximation method to adjust the strength of each segment of the signal to within the preset threshold range. The threshold is preset to ±5% of the signal amplitude, and the segment with optimized signal strength and stability is output.

[0079] S103: Based on the signal strength and stability optimized segments, the specific process of analyzing the stability of each segment of the dynamic electrocardiogram data by comparing the energy distribution difference before and after the adjustment, and generating the electrocardiogram data segment after the stability adjustment is as follows;

[0080] In sub-step S103, based on the signal strength and stability optimized segments, a difference analysis method is adopted, and a comparison diagram of the energy distribution before and after adjustment is drawn using the Matplotlib library. The energy distribution difference is analyzed through the comparison diagram, and the change in stability of each segment is evaluated. By comparing the energy distribution difference before and after adjustment, the difference percentage is used as the stability evaluation standard. The difference percentage calculation is based on the difference between the energy values ​​before and after adjustment divided by the energy value before adjustment. The stability of each segment of dynamic electrocardiogram data is analyzed to generate an electrocardiogram data segment after stability adjustment.

[0081] See also Figure 3 ,For the ECG data segments after stability adjustment, the initial values ​​of the parameters of each segment of dynamic ECG data are set, and the parameters are adjusted according to the characteristics of each segment to reflect the changes of the ECG signals, and the specific steps of generating the parameter update records are as follows;

[0082] S201: for each segment of electrocardiogram data after stability adjustment, a set of initial parameters is set for each segment of data, including basic properties of the signal, including amplitude and frequency response, and the specific process of creating an initial parameter setting record is as follows;

[0083] In sub-step S201, initial parameter setting is performed based on the stability-adjusted ECG data segments. The Python programming language is used in conjunction with the numpy library to set a set of initial parameters for each segment of data, including basic signal properties such as amplitude and frequency response. The amplitude is determined by calculating the average absolute value of the signal within the segment, and the frequency response is determined by analyzing the main frequency components of each signal segment through fast Fourier transform. Through these two steps, an initial parameter setting record is created, which contains detailed information on the amplitude range and main frequency components of each signal segment.

[0084] S202: According to the initial parameter setting record, the characteristics of each segment of data are analyzed, including the change amplitude and frequency characteristics of the signal, and the parameters are adjusted to reflect the characteristics of each segment of the signal. The specific process of forming the parameter modification record is as follows;

[0085] The S202 sub-step is based on the initial parameter setting record and adopts parameter fine-tuning technology. The specific implementation is to visualize the data through the Matplotlib library to intuitively display the characteristics of each segment of data, such as the signal's changing amplitude and frequency characteristics. The parameters are adjusted by modifying the window size and step size in the FFT analysis process to more accurately reflect the characteristics of each segment of the signal. In this process, the adjustment of the frequency characteristics takes into account the changes in the signal's energy distribution in different frequency bands, forming a parameter modification record. The record lists in detail the parameter values ​​after each adjustment, including the new amplitude range and the adjusted frequency response characteristics.

[0086] S203: According to the parameter modification record, referring to the signal variability, the requirement of the specified frequency response and the time window, the parameters of each data segment are adjusted again to reflect the current changes of the ECG signal, and the specific process of obtaining the parameter update record is as follows;

[0087] Sub-step S203 performs signal variability analysis and further adjusts the frequency response based on the parameter modification record. This step uses the signal processing tools in the SciPy library to refer to signal variability indicators, such as the standard deviation and coefficient of variation of the signal, and the requirements for specifying frequency response, such as the increase or decrease in the proportion of energy in a specific frequency band, to make final adjustments to the parameters of each data segment. The adjustment uses dynamic threshold settings to automatically adjust the parameters according to the real-time changes of the ECG signal to ensure that the current state of the ECG signal is reflected, and a parameter update record is generated. The record describes in detail the final state of each data segment after the parameter adjustment, including the final amplitude and frequency response parameters.

[0088] See also Figure 4,Using parameter update records, setting features for various abnormal heart rhythm conditions, analyzing the ,changing trend of ECG signals, and matching signal changes with features, identifying ,potential abnormal heart rhythm, and generating abnormal heart rhythm analysis records ,are the following specific steps;

[0089] S301: Collect parameter update records, define designated signal features for each type of abnormal heart rhythm, reflect typical changes of ECG signals through features, and obtain the abnormal heart rhythm feature definition table in the following specific process;

[0090] Based on the parameter update record, sub-step S301 adopts the data feature analysis method, uses Python language and Pandas library to organize and classify the parameter data, and defines specific signal features for each abnormal heart rhythm, including the waveform variation of the ECG signal, the width of the QRS complex wave, the fluctuation of the heart rate, etc. The typical changes of the ECG signal are reflected by the signal features, and the abnormal heart rhythm feature definition table is obtained, which lists the signal feature parameter ranges corresponding to various abnormal heart rhythms, providing a basis for subsequent matching analysis.

[0091] S302: Based on the abnormal heart rhythm feature definition table, analyzing the change trend of the ECG signal in multiple time windows, combining the local sequence alignment algorithm and the recurrent neural network, matching the change with the feature in the table, identifying the signal change matching the known abnormal heart rhythm feature, and obtaining the abnormal heart rhythm matching information. The specific process is as follows;

[0092] In sub-step S302, based on the definition table of abnormal heart rhythm features, a local sequence alignment algorithm and a recurrent neural network are used to match signal features. First, the Pairwise2 module in the BioPython library is used to perform local sequence alignment. For the specific signal features of each abnormal heart rhythm, similar patterns in the ECG signal are found. Then, the TensorFlow framework is used to build a recurrent neural network model, and the signal data preprocessed by the local sequence alignment algorithm is input. The network model learns the time series features in the ECG signal, matches the changes with the features in the definition table of abnormal heart rhythm features, identifies signal changes that match the known abnormal heart rhythm features, and obtains abnormal heart rhythm matching information, which records in detail the matching abnormal heart rhythm type and its corresponding signal features.

[0093] The local sequence alignment algorithm follows Formula I:

[0094] ;

[0095] The optimal local alignment score between the ECG signal segment and the defined abnormal heart rhythm features is calculated, and the time series characteristics of the ECG signal are analyzed and its long-term dependencies are learned by combining the local sequence alignment results and the learned pattern features to obtain the abnormal heart rhythm matching information.

[0096] in, The sequence alignment score of the reference ECG signal feature is at position and . is the original sequence at position and The maximum alignment score without reference to ECG signal features. For location ECG signal paragraph and location Symptoms of abnormal heart rhythm The matching score between . are the QRS complex widths , heart rate fluctuations , Waveform variation , signal strength The weight coefficient of The weight coefficient is determined by analyzing the influence of QRS complex wave width on the accuracy of abnormal heart rhythm recognition. The weight coefficient is determined by statistically analyzing the correlation strength between heart rate fluctuations and various types of heart rhythm abnormalities. The weight coefficient is determined by evaluating the ability of waveform variability to distinguish different arrhythmias. The weighting factor is determined by comparing the difference in signal intensity between normal and abnormal ECGs. It is the width of the QRS complex wave, reflecting one of the waveform characteristics of the ECG signal. It is the volatility of heart rate, which measures the speed and irregularity of heart rate changes. It is the waveform variability, which reflects the degree of change of the ECG waveform in different time windows. is the signal strength, which represents the average amplitude value of the ECG signal. is the gap penalty value referring to the ECG signal characteristics, used for insertion or deletion operations;

[0097] The execution process is as follows:

[0098] The ECG data is processed by denoising and baseline correction to clearly reflect the characteristics of the ECG signal.

[0099] Extract QRS complex width from ECG signal , use the detection algorithm to find the QRS complex wave and calculate its width. Calculate heart rate fluctuations , measured by analyzing the variability of the intervals between consecutive heartbeats. Determine waveform variability , by comparing the changes in the ECG waveform in different time windows. Measuring signal strength , calculate the average amplitude of the ECG signal.

[0100] Apply formula I to perform local sequence alignment and adjust the matching score by inserting feature weights , and calculate the gap penalty to find the optimal local alignment score between the ECG signal segment and the defined arrhythmia features.

[0101] Using the preprocessed ECG signal data and the results of local sequence alignment as input, the recurrent neural network model is used to further analyze the time series characteristics in the ECG signal, learn its long-term dependencies, and identify more complex abnormal rhythm patterns.

[0102] The analysis results of local sequence alignment and recurrent neural network are integrated to identify ECG signal changes that match known abnormal heart rhythm characteristics and generate abnormal heart rhythm matching information, including the matching abnormal heart rhythm type and its signal characteristics.

[0103] S303: Summarize the contents of the abnormal heart rhythm matching information, analyze the frequency and intensity of each abnormal heart rhythm feature, identify potential abnormal heart rhythm types accordingly, and generate abnormal heart rhythm analysis records. The specific process is as follows:

[0104] Sub-step S303 summarizes the contents of the abnormal heart rhythm matching information, adopts frequency analysis and intensity assessment technology, uses Python and Numpy library for numerical calculation, analyzes the frequency and signal strength of each abnormal heart rhythm feature in the ECG signal, and identifies potential abnormal heart rhythm types through statistical analysis of frequency and quantitative evaluation of signal strength, and generates an abnormal heart rhythm analysis record, which provides a detailed description of each identified abnormal heart rhythm, including the frequency, intensity and possible health risk assessment of the abnormality, providing important diagnostic information for medical professionals.

[0105] See also Figure 5 ,Based on the recognition record of abnormal heart rhythm, the ECG data segment is pattern matched,and compared with the known abnormal heart rhythm pattern, and the abnormal heart rhythm pattern of the same type as the ECG signal is identified by calculating the similarity of each matching result. The specific steps of generating the abnormal heart rhythm pattern similarity analysis result are as follows;

[0106] S401: Using the heart rhythm abnormality analysis record, arranging the heart rhythm abnormality information identified in the electrocardiogram data segment, including the abnormality type and its signal characteristics, and obtaining the heart rhythm abnormality identification information table. The specific process is as follows;

[0107] The S401 sub-step is based on the abnormal rhythm analysis records, adopts the data aggregation method, uses the Python language and the pandas library to process the electrocardiogram data, organizes and summarizes the abnormal rhythm information of each data segment, including the abnormal type and its signal characteristics, and extracts key information such as the amplitude, frequency and duration of the abnormal waveform by traversing each data segment. The information is summarized into an abnormal rhythm identification information table, which records in detail the characteristic parameters of each abnormal rhythm, such as the width of the QRS complex wave and the variability of the heart rate, to obtain the abnormal rhythm identification information table.

[0108] S402: According to the abnormal heart rhythm identification information table, pattern matching is performed on each electrocardiogram data segment, and compared with known abnormal heart rhythm patterns in the database, and the matching degree of each pattern is recorded to obtain the pattern matching degree record table. The specific process is as follows;

[0109] The S402 sub-step is based on the abnormal rhythm identification information table and adopts pattern matching technology. The specific execution is to perform pattern matching on each ECG data segment through Python language combined with the K nearest neighbor algorithm in the scikit-learn library, and compare it with the known abnormal rhythm patterns in the database. In this process, the matching degree of each pattern is calculated by comparing the feature similarity between the ECG data segment and the abnormal rhythm template, including the similarity of the time series and the matching of the morphological features, to obtain a pattern matching record table, which lists the matching scores of each data segment with various known abnormal rhythm patterns, and generates a pattern matching record table.

[0110] S403: Analyze the pattern matching record table, calculate the similarity of each matching result, select the abnormal heart rhythm pattern that matches the ECG signal, and generate the abnormal heart rhythm pattern similarity analysis result. The specific process is as follows;

[0111] Sub-step S403 analyzes the pattern matching record table and adopts a similarity calculation method. Specifically, the matching results are evaluated by using the cosine similarity formula to calculate the similarity score between each data segment and the abnormal heart rhythm template. During this calculation process, the dimension and direction of the ECG signal feature vector are taken into consideration to determine which abnormal heart rhythm pattern the signal is closest to. The abnormal heart rhythm patterns with high matching degrees are screened by setting a similarity threshold. Finally, the abnormal heart rhythm pattern that matches the ECG signal is identified to generate an abnormal heart rhythm pattern similarity analysis result.

[0112] See also Figure 6 ,Using the similarity analysis results of abnormal heart rhythm patterns, analyzing the abnormal heart rhythm patterns with similarity, constructing the feature vector of each abnormal heart rhythm, and identifying the abnormal heart rhythm features in the ECG signal, the specific steps of generating the analysis results of abnormal heart rhythm types are as follows;

[0113] S501: Based on the results of the similarity analysis of abnormal heart rhythm patterns, the abnormal heart rhythm patterns with similarity are screened, the pattern features are classified, and the specific process of obtaining the initial abnormal heart rhythm pattern classification results is as follows;

[0114] In sub-step S501, based on the similarity analysis results of abnormal heart rhythm patterns, a feature classification method is used to screen and classify abnormal heart rhythm patterns with high similarity through Python language combined with NumPy library. In this process, the conditional screening function is used to screen according to the conditions where the similarity score is higher than the preset threshold. The threshold is set to the upper quartile of the similarity score, so as to determine abnormal heart rhythm patterns with significant similarity, and classify and organize the pattern features, including but not limited to waveform features, time intervals and frequency features, to obtain the initial abnormal heart rhythm pattern classification results. The classification results are subdivided based on the type of abnormal heart rhythm, such as ventricular premature beats, atrial premature beats, etc., to generate the initial abnormal heart rhythm pattern classification results.

[0115] S502: Analyze the feature vector of each type of abnormal heart rhythm according to the initial abnormal heart rhythm pattern classification result, including its frequency, amplitude and time characteristics, and obtain the specific process of the feature vector model as follows;

[0116] In sub-step S502, based on the initial abnormal heart rhythm pattern classification results, feature vector construction technology is used to analyze the feature vector of each type of abnormal heart rhythm through Python language and SciPy library, including calculating multi-dimensional values ​​of abnormal heart rhythm characteristics, such as frequency range, maximum amplitude and typical waveform duration, and using parameters to construct a feature vector model. The model defines a multi-dimensional feature space for each abnormal heart rhythm, in which each dimension represents a specific abnormal heart rhythm feature, and generates a feature vector model.

[0117] S503: using the feature vector model, by comparing the similarity between the ECG signal and the feature vector of the abnormal heart rhythm, identifying the feature of the abnormal heart rhythm in the ECG signal, and generating the analysis result of the abnormal heart rhythm type. The specific process is as follows;

[0118] Sub-step S503 uses the feature vector model and similarity comparison technology to compare the similarity between the ECG signal and the abnormal rhythm feature vector through the Python language and the cosine similarity function in the sklearn library. This process refers to the position and direction of the ECG signal and the abnormal rhythm feature vector in the multidimensional feature space, calculates the cosine similarity between them, and identifies the abnormal rhythm features in the ECG signal. The abnormal rhythm features with a similarity higher than another preset threshold (set to 0.75, indicating a higher similarity) are considered to match the current ECG signal, thereby identifying the abnormal rhythm type in the ECG signal and generating an abnormal rhythm type analysis result.

[0119] See also Figure 7,Analyze the analysis results of abnormal heart rhythm types, measure the variability index of ECG signals in each segment, analyze the changing rules of the indexes, reveal the nonlinear characteristics and hidden abnormal patterns in the signal, and generate the signal complexity analysis results in the following specific steps;

[0120] S601: Based on the analysis results of the abnormal heart rhythm type, the ECG data is preprocessed, the box dimension algorithm is used to analyze the complexity and self-similarity of the ECG signal, and the fractal dimension of the signal is referred to, combined with the machine learning algorithm, to identify and classify the abnormal heart rhythm pattern in the ECG signal, and the specific process of obtaining the variability index measurement result is as follows;

[0121] In sub-step S601, based on the analysis results of the types of abnormal heart rhythm, a data preprocessing process is adopted, including denoising and baseline correction. The Python language is combined with the savgol_filter in the SciPy library to implement smoothing filtering and denoising. The detrend method is applied to remove the baseline drift. Then the box dimension algorithm is adopted. Specifically, the box dimension of each ECG signal segment is calculated by calling the numpy library. The signal complexity is estimated by the minimum number of "boxes" covering the ECG signal data points. The complexity and self-similarity of the ECG signal are analyzed. Combined with the machine learning algorithm, specifically the random forest classifier in the scikit-learn library is used to identify and classify the abnormal heart rhythm patterns in the ECG signal according to the fractal dimension of the signal and the preprocessed features, and the variability index measurement results are obtained.

[0122] The box dimension algorithm follows formula II:

[0123] ;

[0124] Calculate the fractal dimension of the ECG signal and generate the complexity and self-similarity values ​​of the ECG signal;

[0125] in, is the fractal dimension of the ECG signal, For scale The minimum number of boxes required to cover the ECG signal. is the side length of the box, representing the observation window at different scales. For heart rate, consider the changes in heartbeat frequency in the ECG signal. To correct the QT interval, the length adjustment of the QT interval in the ECG signal is evaluated and corrected for changes in heart rate. It is the ST segment deviation, which measures the deviation of the ST segment in the ECG signal and identifies myocardial ischemia. P wave duration is used to analyze the duration of atrial activity in the ECG signal. are weight coefficients, corresponding to heart rate , Corrected QT interval , ST segment deviation , P wave duration The importance of reflects the contribution of each parameter to the complexity measurement of ECG signals. The weight coefficient is determined by performing regression analysis on historical electrocardiogram data to evaluate the contribution of QRS complex width in the detection of abnormal heart rhythm. The weight coefficients were determined by analyzing the correlation between heart rate fluctuations and known types of heart rhythm abnormalities using Pearson correlation coefficient analysis. The weight coefficients were determined by analyzing multiple case studies of waveform variation in ECG signals and their role in distinguishing various abnormal heart rhythm conditions using analysis of variance. The weight coefficient is determined by studying the difference in signal strength between normal and abnormal ECGs, applying a decision tree classifier to compare and analyze the results, and selecting the optimal threshold;

[0126] The execution process is as follows:

[0127] The ECG data were denoised and baseline corrected.

[0128] The heart rate (HR) was extracted from the preprocessed ECG signal and the number of heart beats per minute was calculated.

[0129] The corrected QT interval (QTc) was calculated using the Bazett formula to account for the effect of heart rate variability on the QT interval.

[0130] Measurement of ST segment deviation ( ), identify and quantify the degree of ST segment deviation in ECG signals.

[0131] Determine the duration of the P wave ( ), measures the duration of the P wave in the ECG signal.

[0132] Calculate the box dimension of each ECG signal segment according to Formula II, including calculating the minimum number of boxes at different scales , and obtain the fractal dimension of the ECG signal.

[0133] Combined with machine learning models to analyze signal complexity and self-similarity, it can identify and classify abnormal rhythm patterns in ECG signals.

[0134] The identified abnormal heart rhythm types and their associated characteristics are summarized into a report and provided to medical professionals for further analysis and diagnostic decision-making.

[0135] S602: Iteratively analyzing the variability index measurement results, analyzing the changing rules of the ECG signal variability index, identifying the nonlinear characteristics of the signal, and obtaining the specific process of the nonlinear characteristic analysis results is as follows;

[0136] The S602 sub-step iteratively analyzes the variability index measurement results, using the variability index analysis technology, specifically by calculating the standard deviation and mean in each segment of the ECG signal, using the Python language combined with the numpy library for numerical operations, so as to identify the nonlinear characteristics of the ECG signal, analyze the changing rules of the ECG signal variability index, and reveal the nonlinear characteristics and variability patterns in the ECG signal by drawing a trend graph of the variability index over time and using the matplotlib library for visualization, and obtain the nonlinear characteristic analysis results.

[0137] S603: According to the nonlinear feature analysis results, the hidden abnormal pattern in the ECG signal is revealed, and the relationship between the pattern and the known abnormal heart rhythm is determined through comparative analysis. The specific process of generating the signal complexity analysis result is as follows;

[0138] Sub-step S603 uses a pattern comparison analysis method based on the nonlinear feature analysis results. Specifically, by comparing the nonlinear features of the ECG signal with the known nonlinear abnormal patterns in the database, the Python language is combined with similarity calculation methods in the scikit-learn library, such as cosine similarity, to evaluate the similarity between the ECG signal and various abnormal heart rhythm patterns, reveal the hidden abnormal patterns in the ECG signal, determine the relationship between the pattern and the known abnormal heart rhythm through comparative analysis, and generate signal complexity analysis results.

[0139] See also Figure 8 , a dynamic electrocardiogram data processing system, the system comprising:

[0140] The data segmentation module divides the dynamic electrocardiogram data into several data segments according to the time series or the number of data points. The amount of data included in each segment can be independently analyzed to obtain a data segment set;

[0141] The stability analysis module performs signal strength adjustment on each segment in the data segment set, calculates energy distribution, and makes stability judgment on each segment to generate stability analysis results;

[0142] The parameter adjustment module sets the initial parameters for each data segment according to the stability analysis results, and adjusts the parameters according to the signal characteristics of each segment to reflect the changing characteristics of the ECG signal and obtain the parameter update record;

[0143] The heart rhythm abnormality feature matching module uses parameter update records to analyze changes in ECG signals for defined heart rhythm abnormality features, matches the changes with the features, identifies matching heart rhythm abnormalities, and constructs heart rhythm abnormality analysis records;

[0144] The pattern similarity analysis module performs a comparison between the ECG data segment and the known abnormal heart rhythm pattern based on the abnormal heart rhythm analysis record, calculates the matching similarity, screens the abnormal heart rhythm pattern that matches the ECG signal, and generates a similarity analysis result;

[0145] The feature vector construction module analyzes the abnormal heart rhythm patterns selected in the similarity analysis results, summarizes the feature vectors of each abnormal heart rhythm, including multiple characteristics such as frequency, amplitude and time, and obtains a set of abnormal heart rhythm feature vectors;

[0146] The signal complexity analysis module measures the variability index of the ECG signal in each segment based on the abnormal rhythm feature vector set, analyzes the variation pattern, reveals the nonlinear characteristics and potential abnormal patterns in the signal, and establishes the signal complexity analysis results.

[0147] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for processing dynamic electrocardiogram data, characterized in that: The following steps are involved: Segment the acquired dynamic electrocardiogram data, adjust the signal strength of each segment, calculate the energy distribution of each segment, and determine its stability, so as to generate electrocardiogram data segments after stability adjustment; For the ECG data segments after stability adjustment, set the initial value of the parameters of each segment of dynamic ECG data, and adjust the parameters according to the characteristics of each segment to reflect the changes of the ECG signal and generate a parameter update record; Using the parameter to update the record, setting features for various abnormal heart rhythm conditions, analyzing the trend of the ECG signal, and matching the signal changes with the features, identifying potential abnormal heart rhythms, and generating abnormal heart rhythm analysis records; Based on the recognition record of abnormal heart rhythm, pattern matching is performed on the electrocardiogram data segment, and it is compared with the known abnormal heart rhythm pattern, and the abnormal heart rhythm pattern similar to the electrocardiogram signal is identified by calculating the similarity of each matching result, and the abnormal heart rhythm pattern similarity analysis result is generated; Analyze the abnormal heart rhythm patterns with similarity using the abnormal heart rhythm pattern similarity analysis results, construct feature vectors for each abnormal heart rhythm, identify abnormal heart rhythm features in the electrocardiogram signal, and generate abnormal heart rhythm type analysis results; The analysis results of the abnormal heart rhythm types are analyzed, the variability index of the electrocardiogram signal in each segment is measured, the changing rules of the index are analyzed, the nonlinear characteristics and hidden abnormal patterns in the signal are revealed, and the signal complexity analysis results are generated.

2. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: The stability-adjusted ECG data segments include adjusted signal amplitudes, signal stability ratings, and energy distribution characteristics of each segment; the parameter update records include adjusted time window size, signal frequency response adjustment values, and signal variability adjustment indexes; the abnormal heart rhythm analysis records include abnormal heart rhythm prediction probabilities, abnormal feature strength judgments, and abnormal trend indicators; the abnormal heart rhythm pattern similarity analysis results include abnormal heart rhythm matching scores, matching pattern identifiers, and pattern similarity levels; the abnormal heart rhythm type analysis results include abnormal heart rhythm type confirmation, abnormal feature vector matching, and abnormal category identification confidence; the signal complexity analysis results include signal variability index details, segment nonlinear feature overviews, and potential abnormal pattern signal characteristics.

3. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: The acquired dynamic electrocardiogram data is segmented, the signal strength of each segment is adjusted, the energy distribution of each segment is calculated, and its stability is determined. The specific steps for generating the electrocardiogram data segment after stability adjustment are as follows: The acquired dynamic electrocardiogram data is segmented, the signal strength is adjusted segment by segment, the energy distribution of each segment is calculated, and the adjusted signal segment is obtained; Analyze each of the adjusted signal segments, determine the stability of the signal peak value, mean value, and energy distribution, iteratively adjust the signal strength, and output a segment with optimized signal strength and stability; Based on the signal strength and stability optimized segments, the stability of each segment of dynamic electrocardiogram data is analyzed by comparing the energy distribution difference before and after adjustment, and the electrocardiogram data segment after stability adjustment is generated.

4. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: For the ECG data segment after stability adjustment, the initial value of the parameter of each segment of dynamic ECG data is set, and the parameter is adjusted according to the characteristics of each segment to reflect the change of the ECG signal. The specific steps of generating the parameter update record are as follows: For the stability-adjusted ECG data segments, set a set of initial parameters for each segment of data, including basic properties of the signal, including amplitude and frequency response, and create an initial parameter setting record; According to the initial parameter setting record, the characteristics of each segment of data are analyzed, including the change amplitude and frequency characteristics of the signal, and the parameters are adjusted to reflect the characteristics of each segment of the signal, thereby forming a parameter modification record; According to the parameter modification record, referring to the signal variability, the requirement of the specified frequency response and the time window, the parameters of each data segment are readjusted to reflect the current changes of the ECG signal, and the parameter update record is obtained.

5. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: The specific steps of using the parameter update record, setting features for various abnormal heart rhythm conditions, analyzing the change trend of the ECG signal, matching the signal change with the features, identifying potential abnormal heart rhythm, and generating abnormal heart rhythm analysis records are as follows: Collect the parameter update records, define a specified signal feature for each abnormal heart rhythm, reflect the typical changes of the electrocardiogram signal through the feature, and obtain a definition table of abnormal heart rhythm features; Based on the abnormal heart rhythm feature definition table, analyzing the changing trends of the ECG signal in multiple time windows, combining the local sequence alignment algorithm and the recurrent neural network, matching the changes with the features in the table, identifying the signal changes that match the known abnormal heart rhythm features, and obtaining abnormal heart rhythm matching information; The local sequence alignment algorithm is according to Formula I: ； The optimal local alignment score between the ECG signal segment and the defined abnormal heart rhythm features is calculated, and the time series characteristics of the ECG signal are analyzed and its long-term dependencies are learned by combining the local sequence alignment results and the learned pattern features to obtain the abnormal heart rhythm matching information. in, The sequence alignment score of the reference ECG signal feature is at position and , is the original sequence at position and The maximum alignment score of , without reference to ECG signal features, For location ECG signal paragraph and location Symptoms of abnormal heart rhythm The matching score between are the QRS complex widths , heart rate fluctuations , Waveform variation , signal strength The weight coefficient of The weight coefficient is determined by analyzing the influence of QRS complex wave width on the accuracy of abnormal heart rhythm recognition. The weight coefficient is determined by statistically analyzing the correlation strength between heart rate fluctuations and each type of heart rhythm abnormality. The weight coefficient is determined by evaluating the ability of waveform variability to distinguish between various abnormal heart rhythms. The weight coefficient is determined by comparing the difference in signal strength between normal ECG and abnormal ECG. is the width of the QRS complex, The fluctuation of heart rate, is the waveform variability, is the signal strength, is the gap penalty value referring to the ECG signal characteristics; The contents of the abnormal heart rhythm matching information are summarized, and the frequency and intensity of each abnormal heart rhythm feature are analyzed to identify potential abnormal heart rhythm types and generate abnormal heart rhythm analysis records.

6. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: Based on the recognition record of the abnormal heart rhythm, pattern matching is performed on the electrocardiogram data segment, and it is compared with the known abnormal heart rhythm pattern. By calculating the similarity of each matching result, the abnormal heart rhythm pattern similar to the electrocardiogram signal is identified, and the specific steps of generating the abnormal heart rhythm pattern similarity analysis result are as follows: Using the abnormal heart rhythm analysis record, sorting the abnormal heart rhythm information identified in the electrocardiogram data segment, including the abnormal type and its signal characteristics, to obtain an abnormal heart rhythm identification information table; According to the abnormal heart rhythm identification information table, pattern matching is performed on each electrocardiogram data segment, and compared with known abnormal heart rhythm patterns in the database, and the matching degree of each pattern is recorded to obtain a pattern matching degree record table; The pattern matching record table is analyzed, the similarity of each matching result is calculated, the abnormal heart rhythm pattern matching the electrocardiogram signal is screened, and the abnormal heart rhythm pattern similarity analysis result is generated.

7. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: The specific steps of using the abnormal heart rhythm pattern similarity analysis result to analyze the abnormal heart rhythm patterns with similarity, constructing the feature vector of each abnormal heart rhythm, and identifying the abnormal heart rhythm features in the electrocardiogram signal to generate the abnormal heart rhythm type analysis result are as follows: Based on the similarity analysis result of the abnormal heart rhythm pattern, screening abnormal heart rhythm patterns with similarity, classifying the pattern features, and obtaining an initial abnormal heart rhythm pattern classification result; Analyze the characteristic vector of each type of abnormal heart rhythm according to the initial abnormal heart rhythm pattern classification result, including multiple factors such as frequency, amplitude and time characteristics, to obtain a characteristic vector model; By using the feature vector model, the feature of abnormal heart rhythm in the electrocardiogram is identified by comparing the similarity between the electrocardiogram signal and the feature vector of abnormal heart rhythm, and the analysis result of the abnormal heart rhythm type is generated.

8. The method for processing dynamic electrocardiogram data according to claim 1, characterized in that: The specific steps of analyzing the analysis results of the abnormal heart rhythm type, measuring the variability index of the ECG signal in each segment, analyzing the change law of the index, revealing the nonlinear characteristics and hidden abnormal patterns in the signal, and generating the signal complexity analysis results are as follows: Based on the analysis results of the abnormal heart rhythm types, the electrocardiogram data is preprocessed, and the complexity and self-similarity of the electrocardiogram signal are analyzed by using a box dimension algorithm, and the abnormal heart rhythm patterns in the electrocardiogram signal are identified and classified by referring to the fractal dimension of the signal and combining a machine learning algorithm to obtain a variability index measurement result; The box dimension algorithm is according to Formula II: ； Calculate the fractal dimension of the ECG signal and generate the complexity and self-similarity values ​​of the ECG signal; in, is the fractal dimension of the ECG signal, For scale The minimum number of boxes required to cover the ECG signal is: is the side length of the box, is the heart rate, To correct the QT interval, evaluate the length adjustment of the QT interval in the ECG signal, For ST segment deviation, is the P wave duration, are weight coefficients, corresponding to heart rate , Corrected QT interval , ST segment deviation , P wave duration The importance of The weight coefficient is determined by performing regression analysis on historical ECG data to evaluate the contribution of QRS complex width in abnormal heart rhythm detection. The weight coefficients were determined by analyzing the correlation between heart rate fluctuations and known types of abnormal heart rhythms using Pearson correlation coefficient analysis. The weight coefficients were determined by using variance analysis through multiple case studies of waveform variation in ECG signals to analyze their role in distinguishing various abnormal heart rhythm conditions. The weight coefficient is determined by studying the difference in signal strength between normal and abnormal ECGs, applying a decision tree classifier to compare and analyze the results, and selecting the optimal threshold; Iteratively analyzing the variability index measurement results, analyzing the changing rules of the ECG signal variability index, identifying the nonlinear characteristics of the signal, and obtaining nonlinear characteristic analysis results; According to the nonlinear characteristic analysis results, the hidden abnormal pattern in the ECG signal is revealed, and through comparative analysis, the relationship between the pattern and the known abnormal heart rhythm is determined to generate a signal complexity analysis result.

9. A dynamic electrocardiogram data processing system, characterized in that: According to any one of claims 1 to 8, the system is implemented as a method for processing dynamic electrocardiogram data, comprising: The data segmentation module divides the dynamic electrocardiogram data into several data segments according to the time series or the number of data points. The amount of data included in each segment can be independently analyzed to obtain a data segment set; The stability analysis module performs signal strength adjustment on each segment in the data segment set, calculates energy distribution, and makes stability judgment on each segment to generate a stability analysis result; The parameter adjustment module sets initial parameters for each data segment according to the stability analysis results, and adjusts the parameters according to the signal characteristics of each segment to reflect the changing characteristics of the ECG signal and obtain a parameter update record; The heart rhythm abnormality feature matching module uses the parameter to update the record, analyzes the changes in the ECG signal for the defined heart rhythm abnormality features, matches the changes with the features, identifies the matched heart rhythm abnormality, and constructs the heart rhythm abnormality analysis record; The pattern similarity analysis module compares the electrocardiogram data segment with the known abnormal heart rhythm pattern based on the abnormal heart rhythm analysis record, calculates the similarity of the match, screens the abnormal heart rhythm pattern that matches the electrocardiogram signal, and generates a similarity analysis result; The feature vector construction module analyzes the abnormal heart rhythm patterns selected in the similarity analysis results, summarizes the feature vectors of each abnormal heart rhythm, including multiple characteristics such as frequency, amplitude and time, and obtains a set of abnormal heart rhythm feature vectors; The signal complexity analysis module measures the variability index of the ECG signal in each segment based on the abnormal rhythm feature vector set, analyzes the variation pattern, reveals the nonlinear characteristics and potential abnormal patterns in the signal, and establishes the signal complexity analysis results.

Images