Atrial fibrillation signal identification method and apparatus, electronic device, and computer readable medium

Abstract

Embodiments of the present disclosure disclose a method and device for atrial fibrillation signal identification, an electronic device and a computer readable medium. A specific implementation of the method comprises: acquiring an electrocardiogram signal through an electrocardiogram acquisition device; performing noise reduction processing on the electrocardiogram signal to generate a noise-reduced electrocardiogram signal; locating RR intervals in the noise-reduced electrocardiogram signal to generate a sequence of RR intervals to be marked; identifying RR intervals affected by body position changes and regular arrhythmia through XGboost and random deep forest model on the sequence of RR intervals to be marked to generate an atrial fibrillation event group; performing morphological analysis on the noise-reduced electrocardiogram signal to generate electrocardiogram P wave features; dynamically monitoring the electrocardiogram P wave features, and adjusting the starting time point and the ending time point of each atrial fibrillation event in the atrial fibrillation event group according to the result of dynamic monitoring to obtain an optimized atrial fibrillation signal group. The implementation can improve the accuracy of atrial fibrillation signal identification.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the field of signal recognition, and in particular, to an atrial fibrillation signal recognition method and device, an electronic device, and a computer readable medium. BACKGROUND

[0002] Atrial fibrillation (AF) is one of the most common arrhythmias in clinical practice, characterized by highly irregular atrial electrical activity, accompanied by atrial dysfunction and arrhythmia. AF not only affects the quality of life of patients, but also significantly increases the risk of stroke, heart failure, and cardiovascular disease mortality. According to its duration, atrial fibrillation can be divided into paroxysmal atrial fibrillation (PAF), persistent atrial fibrillation, and permanent atrial fibrillation. Among them, paroxysmal atrial fibrillation (PAF) has the characteristics of intermittent attack, and usually recovers on its own within a few minutes to a few hours, bringing great challenges to clinical diagnosis.

[0003] In the early stage of PAF, patients may have no obvious symptoms, or only show non-specific symptoms such as occasional palpitations, chest tightness, dizziness, etc. Due to the short duration of PAF, it is difficult to capture through conventional 12-lead electrocardiogram (ECG). As a tool that can record electrocardiogram signals for 24 hours or even longer, Holter monitoring has become an important means for the diagnosis of PAF. However, due to the randomness and transience of PAF attacks, even with Holter monitoring, it may not be possible to record typical AF waveforms due to the absence of atrial fibrillation or the extremely short duration of AF, resulting in missed diagnosis. In addition, factors such as patient daily activities, body position changes, and poor electrode contact may introduce noise, further affecting the accurate identification of PAF events.

[0004] Currently, the diagnosis of atrial fibrillation still mainly relies on manual electrocardiogram analysis, and doctors identify AF events by visually inspecting electrocardiogram waveforms. Although this method has high reliability, it has the following limitations:

[0005] 1. Dependence on expert experience: Traditional manual interpretation is greatly influenced by subjective factors, and the level of doctor experience and film review directly determines the accuracy of diagnosis.

[0006] 2. High time cost: Dynamic electrocardiogram recording time usually lasts for 24-72 hours, and doctors need to analyze a large amount of electrocardiogram data frame by frame, resulting in a huge amount of diagnostic work.

[0007] 3. Difficulty in identifying short-onset AF: For short AF events with a duration of less than 1 minute, it is more difficult for artificial identification, and it is easy to cause missed diagnosis or misdiagnosis due to heart rate variability or artifact interference.

[0008] The above information disclosed in this Background section is only for the purpose of enhancing the understanding of the background of the inventive concepts, and therefore, it can contain information that does not form the prior art that is already known to those of ordinary skill in the art in the country. SUMMARY

[0009] This section provides a summary of the concepts, which will be described in detail in the following detailed description. This summary section is not intended to identify key or essential features of the claimed technology nor is it intended to be used to limit the scope of the claimed technology.

[0010] Some embodiments of the present disclosure propose an atrial fibrillation signal identification method, device, electronic equipment and computer readable medium to solve one or more of the technical problems mentioned in the background section.

[0011] In a first aspect, some embodiments of the present disclosure provide an atrial fibrillation signal identification method, which comprises: acquiring an electrocardiogram signal through an electrocardiogram acquisition device; performing noise reduction processing on the electrocardiogram signal to generate a noise-reduced electrocardiogram signal, wherein the R-wave peak signal position in the noise-reduced electrocardiogram signal is marked by R-point detection; locating the RR interval in the noise-reduced electrocardiogram signal to generate a sequence of RR intervals to be marked, wherein each RR interval to be marked in the sequence of RR intervals to be marked corresponds to a group of atrial fibrillation start time points and atrial fibrillation end time points; identifying the RR interval sequence of body position change interference and regular arrhythmia through XGboost and random deep forest model on the sequence of RR intervals to be marked, and excluding the RR interval sequence of these regions from the identification region to complete purification processing, and performing atrial fibrillation event analysis on the sequence of RR intervals to be marked after purification processing to generate an atrial fibrillation event group, wherein the variability of each RR interval to be marked in the sequence of RR intervals to be marked after purification processing is dynamically monitored through a sliding window; performing morphological analysis on the noise-reduced electrocardiogram signal to generate electrocardiogram P-wave features, wherein the electrocardiogram P-wave features include P-wave start point, P-wave peak value and P-wave end point; dynamically monitoring the electrocardiogram P-wave features, and adjusting the atrial fibrillation start time point and the atrial fibrillation end time point of each atrial fibrillation event in the atrial fibrillation event group according to the result of dynamic monitoring to obtain an optimized atrial fibrillation signal group.

[0012] In a second aspect, some embodiments of the present disclosure provide an atrial fibrillation signal identification device, comprising: an acquisition unit configured to acquire an electrocardiogram signal through an electrocardio acquisition device; a noise reduction processing unit configured to perform noise reduction processing on the electrocardiogram signal to generate a noise-reduced electrocardiogram signal, wherein R-wave peak signal positions in the noise-reduced electrocardiogram signal are marked by R-point detection; a signal positioning unit configured to position RR intervals in the noise-reduced electrocardiogram signal to generate a sequence of RR intervals to be marked, wherein each RR interval to be marked in the sequence of RR intervals to be marked corresponds to a group of atrial fibrillation start time points and atrial fibrillation end time points; an atrial fibrillation event analysis unit configured to identify body position change interference and regular arrhythmia RR intervals in the sequence of RR intervals to be marked through XGboost and random deep forest model, exclude RR interval sequences in these regions from the identified regions to complete purification processing, and perform atrial fibrillation event analysis on the sequence of RR intervals to be marked after purification processing to generate an atrial fibrillation event group, wherein the variability of each RR interval to be marked in the sequence of RR intervals to be marked after purification processing is dynamically monitored through a sliding window; a morphological analysis unit configured to perform morphological analysis on the noise-reduced electrocardiogram signal to generate electrocardiogram P wave features, wherein the electrocardiogram P wave features include P wave start points, P wave peaks, and P wave end points; a dynamic monitoring and optimization unit configured to dynamically monitor the electrocardiogram P wave features, and adjust the atrial fibrillation start time points and the atrial fibrillation end time points of each atrial fibrillation event in the atrial fibrillation event group according to the results of dynamic monitoring to obtain an optimized atrial fibrillation signal group.

[0013] In a third aspect, some embodiments of the present disclosure provide an electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any of the implementations of the first aspect.

[0014] In a fourth aspect, some embodiments of the present disclosure provide a computer readable medium having a computer program stored thereon, wherein the program is executed by a processor to implement the method described in any of the implementations of the first aspect.

[0015] The various embodiments of this disclosure have the following beneficial effects: the atrial fibrillation signal identification method of some embodiments of this disclosure can improve the accuracy of atrial fibrillation event identification. Specifically, the reason why it is easy to miss or misdiagnose due to arrhythmia or artifact interference is that: traditional manual interpretation methods are greatly affected by subjective factors, the doctor's experience and the level of review of the film directly determine the accuracy of the diagnosis, and for brief AF events lasting less than 1 minute, manual identification is more difficult. Based on this, the atrial fibrillation signal identification method of some embodiments of this disclosure firstly acquires an electrocardiogram (ECG) signal through an ECG acquisition device. Noise is then removed to ensure signal clarity. Secondly, the ECG signal is denoised to generate a denoised ECG signal. Among them, the position of the R-wave peak signal in the denoised ECG signal is marked by R-point detection. The R-point marking can provide basic data for subsequent RR interval calculation. Then, the RR interval in the denoised ECG signal is located to generate a RR interval sequence to be marked. In this study, each RR interval in the aforementioned unlabeled RR interval sequence corresponds to a set of atrial fibrillation onset and termination time points. Preliminary RR interval localization helps determine usable RR intervals and reduces data interference. Subsequently, the unlabeled RR interval sequence is used to identify RR intervals affected by postural changes and regular arrhythmias using XGboost and the Stochastic Forest model. RR interval sequences from these regions are excluded from the identification area, completing the purification process. Atrial fibrillation event analysis is then performed on the purified unlabeled RR interval sequence to generate atrial fibrillation event sets. A sliding window is used to dynamically monitor the variability of each unlabeled RR interval in the purified unlabeled RR interval sequence. This dynamic monitoring helps to preliminarily locate the occurrence intervals of atrial fibrillation. Next, morphological analysis is performed on the noise-reduced electrocardiogram (ECG) signal to generate ECG P-wave features. These ECG P-wave features include the P-wave onset, P-wave peak, and P-wave end. Morphological analysis further refines the P-wave features. Finally, the aforementioned ECG P-wave characteristics were dynamically monitored, and based on the monitoring results, the atrial fibrillation onset and termination times of each atrial fibrillation event in the atrial fibrillation event group were adjusted to obtain an optimized atrial fibrillation signal group. Here, through more precise P-wave characteristics and dynamic monitoring, atrial fibrillation events can be located more accurately, and the corresponding time points can be optimized. This improves the accuracy of atrial fibrillation signal identification. Attached Figure Description

[0016] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0017] Figure 1 This is a flowchart of some embodiments of the atrial fibrillation signal identification method according to the present disclosure;

[0018] Figure 2 This is a schematic diagram of the marked interference area;

[0019] Figure 3 This is a schematic diagram marking the intervals of atrial fibrillation events;

[0020] Figure 4 This is a schematic diagram illustrating the optimized timing of atrial fibrillation.

[0021] Figure 5 This is a schematic diagram of the structure of some embodiments of the atrial fibrillation signal identification device according to the present disclosure;

[0022] Figure 6 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0023] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0024] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0025] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0026] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0027] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0028] Currently, atrial fibrillation (AF) identification is primarily based on changes in the RR interval. Common methods include:

[0029] 1. Threshold method: A fixed threshold is set for RR interval variability, such as the root mean square difference (RMSSD) exceeding a certain set value, which is then judged as AF. However, due to significant differences in heart rate variability among different individuals, this method may lead to a high false positive rate (oversensitivity) or a high false negative rate (under-identification).

[0030] 2. Pattern matching algorithm: This method uses pre-defined feature templates to match actual ECG data, improving the robustness of arrhythmia (AF) identification. However, this method is easily affected by the diversity of arrhythmia types and has poor generalization ability.

[0031] 3. Artificial intelligence deep learning methods: such as convolutional neural networks, long short-term memory networks and their combined models.

[0032] However, the above method has the following drawbacks: slow inference speed: due to the large amount of dynamic electrocardiogram signal, the computational load of using complex deep neural networks (such as CNN-LSTM combined model) for full-time data analysis is large, the real-time performance is poor, and it is difficult to meet the needs of clinical immediate diagnosis. A 24-hour dynamic electrocardiogram may take several minutes or even tens of minutes to perform atrial fibrillation analysis.

[0033] High data dependence: The training of deep learning models is highly dependent on large-scale labeled data, and the distribution differences between different datasets (such as device sampling rate, signal quality, patient characteristics, etc.) may affect the generalization ability of the model.

[0034] Insufficient ability to identify short-duration AF: Due to the large individual differences in the performance of AF during the RR interval, traditional deep learning methods perform poorly in identifying short-duration AF and are easily affected by artifacts and heart rate variability.

[0035] Furthermore, while deep learning models can learn complex electrocardiogram signal features through training on massive amounts of data, their "black box" nature results in poor interpretability of diagnostic results. Doctors may find it difficult to understand the model's decision-making rationale, potentially hindering its widespread adoption in medical applications. Additionally, some deep learning methods have high computational resource requirements, making them unsuitable for resource-constrained wearable devices or remote medical monitoring systems, further limiting their application in clinical practice.

[0036] Therefore, there is an urgent clinical need for an efficient, accurate, and real-time method for atrial fibrillation (AF) identification to overcome the shortcomings of existing technologies. This invention targets the absolute irregularity of the RR interval and the disappearance or variation of the P wave, combining a rule engine, statistical analysis, and machine learning (RF (Random Forest) + XGBoost) to provide an AF identification scheme with high accuracy, ultra-high speed, real-time performance, and low computational cost.

[0037] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0038] Figure 1 A flow 100 of some embodiments of an atrial fibrillation signal identification method according to the present disclosure is shown. The atrial fibrillation signal identification method includes the following steps:

[0039] Step 101: Obtain electrocardiogram (ECG) signals using an ECG acquisition device.

[0040] In some embodiments, the entity executing the atrial fibrillation signal identification method (e.g., a computing device) can acquire electrocardiogram (ECG) signals via a wired or wireless ECG acquisition device. The ECG signals can be data collected by the ECG acquisition device from a single patient.

[0041] Step 102: Denoise the electrocardiogram signal to generate a denoised electrocardiogram signal.

[0042] In some embodiments, the execution entity may perform noise reduction processing on the electrocardiogram (ECG) signal to generate a noise-reduced ECG signal. The location of the R-wave peak signal in the noise-reduced ECG signal can be marked using R-point detection.

[0043] In some optional implementations of certain embodiments, the execution entity performs noise reduction processing on the electrocardiogram signal to generate a noise-reduced electrocardiogram signal, including:

[0044] The first step is to use bandpass filtering to remove baseline drift and electromyographic interference from the above electrocardiogram signal to obtain a noise-reduced electrocardiogram signal.

[0045] The second step involves detecting the R-wave peak in the denoised ECG signal using adaptive threshold detection and the second derivative method to pinpoint the location of the R-wave peak. Here, the R-point can be detected using adaptive threshold detection and the second derivative method.

[0046] Step 103: Locate the RR intervals in the noise-reduced electrocardiogram signal to generate a sequence of RR intervals to be labeled.

[0047] In some embodiments, the execution entity can locate the RR intervals in the noise-reduced electrocardiogram signal to generate a sequence of RR intervals to be labeled. Each RR interval in the sequence can correspond to a set of atrial fibrillation onset and termination time points. Specifically, the RR interval refers to the interval between two adjacent QRS complex waves in the electrocardiogram signal.

[0048] In some optional implementations of certain embodiments, the execution entity locates the RR intervals in the denoised electrocardiogram signal to generate a sequence of RR intervals to be labeled, including:

[0049] After denoising the ECG signal, obvious abnormalities were removed to obtain the RR interval sequence to be labeled. The abnormal RR variation threshold can be obtained based on the mean ± 3 times the standard deviation to remove obvious abnormalities. Here, the mean and standard deviation can be the mean and standard deviation of each RR interval. The abnormal RR variation threshold can be a range value.

[0050] Step 104: The RR interval sequences to be labeled are identified by XGboost and the random forest model to identify RR intervals that are affected by postural changes and regular arrhythmias, and the RR interval sequences in these regions are excluded from the identification area to complete the purification process. Atrial fibrillation event analysis is then performed on the RR interval sequences to be labeled to generate atrial fibrillation event groups.

[0051] In some embodiments, the aforementioned execution entity can use XGboost and the Random Forest model to identify RR intervals affected by postural changes and regular arrhythmias in the RR interval sequence to be labeled, and exclude RR interval sequences from these regions to complete the purification process. Furthermore, it can perform atrial fibrillation event analysis on the RR interval sequence to be labeled to generate atrial fibrillation event sets. The variability of the RR interval to be labeled can be dynamically calculated using a preset sliding window (e.g., a window size of 15 heartbeats) to avoid the limitations imposed by a fixed time window.

[0052] In some optional implementations of certain embodiments, the execution entity identifies RR intervals in the unlabeled RR interval sequence using XGboost and the Random Forest model to detect postural disturbances and regular arrhythmias, excludes RR interval sequences in these regions from the identification area to complete the purification process, and performs atrial fibrillation event analysis on the purified unlabeled RR interval sequence to generate an atrial fibrillation event set, including:

[0053] The first step involves inputting the aforementioned RR interval sequence to be labeled into a pre-trained random forest model to obtain a preliminary RR interval sequence. This random forest model is used to identify the trend of RR interval changes based on the smoothness of the RR interval sequence to determine if there are any sudden abnormal signals, allowing for the labeling and removal of RR intervals that interfere with postural changes or regular arrhythmias. Specifically, the standard deviation of the removed ECG signal and the corresponding RR intervals (as the trend of RR interval changes) can be input into the pre-trained random forest model to obtain the preliminary RR interval sequence.

[0054] Specifically, a pre-set method for detecting regular arrhythmias can be used to determine the trend of RR interval changes in order to identify whether there are sudden abnormalities (such as premature beats).

[0055] In practice, methods for detecting regular arrhythmias can include: 1. If the RR interval shows a fixed multiple of variation (e.g., bigeminy, trigeminy), it may be a premature beat and needs to be excluded. 2. If the RR variation is highly regular and appears periodically, it is classified as a regular arrhythmia and is not considered for arrhythmia identification. The processing logic here can be: if it conforms to a regular arrhythmia pattern (premature beat), then mark and exclude the data from subsequent analysis. If the RR variation pattern does not show obvious regularity, then proceed to XGBoost for fine classification.

[0056] As an example, see Figure 2 Based on effective data purification of RR intervals, regular RR interval changes caused by premature beats, bigeminy, trigeminy, etc., as well as RR recognition errors caused by electrode detachment, electromyographic interference, etc.

[0057] In addition, the above-mentioned random forest model can be generated by training through the following steps:

[0058] A locally configured atrial fibrillation (AF) case database is used, including noise-interference data, normal heart rhythms (including AF), and regular arrhythmias (such as premature beats) on ECG signals, all annotated by ECG experts. Data preprocessing includes extracting the RR interval from the ECG signals (calculating the time interval between adjacent R waves). Features are calculated: RR interval mutation points (mean ± 3 standard deviations), RR interval variation trends (smoothness, variation patterns), and regular arrhythmia characteristics (fixed periodic variations such as bigeminy and trigeminy). Model input can include classification labels (noise labels, normal signal labels, regular arrhythmia labels, etc.). After training, preliminary classification is performed to remove noise and regular arrhythmia data. The remaining data is then processed by the XGBoost model for fine-grained classification, identifying RR interval variations related to AF.

[0059] The second step involves inputting the RR interval sequence after the initial detection into a pre-trained machine learning model to obtain the RR interval sequence to be labeled. The machine learning model can be used to remove RR intervals after the initial detection that indicate electrode detachment or EMG interference, while the RR intervals to be labeled represent highly variable RR intervals. Here, the machine learning model can be a pre-trained XGBoost model. This machine learning model can remove RR intervals after the initial detection that indicate electrode detachment or EMG interference through electrode detachment detection and EMG interference detection.

[0060] Specifically, electrode detachment detection can be achieved by identifying erroneous RR interval recordings caused by abnormal amplification of continuous RR interval variability. Electromyographic interference (EMG) detection can be performed if RR interval variability is excessively large, but the RR interval variation pattern is highly random; this may indicate EMG interference. This can be determined by combining the noise-to-power ratio in the ECG signal.

[0061] For example, if the machine learning model determines that the RR interval is due to electrode detachment or electromyography interference, then that RR interval segment is removed and not included in AF identification. If the machine learning model determines that the RR interval is a true high-variability RR interval segment, then the data is retained and proceeds to the next step of calculating RR variability using a sliding window.

[0062] Furthermore, misjudgments by the machine learning model can be corrected: the self-similarity of the RR interval variation is calculated. If the variation pattern is chaotic, the data is retained and entered into a sliding window analysis. If the RR interval variation has a fixed pattern (e.g., isochronous mutations), it is still treated as a regular arrhythmia and removed. This ensures that arrhythmia fragments are not mistakenly deleted due to regularity errors, thereby improving the sensitivity of arrhythmia identification.

[0063] As an example, self-similarity can be calculated using the following steps:

[0064] RR self-similarity is primarily used to distinguish regular arrhythmias (such as PVC and PAC) from irregular RR variants associated with arrhythmias (AF). The core idea of ​​self-similarity is to analyze the pattern stability of RR interval sequences. If a segment of RR variants exhibits a periodic or fixed pattern, it is likely a regular arrhythmia rather than AF. Here, we can construct an adjacent RR correlation matrix for the RR interval sequences, assuming the RR interval sequence is:

[0065] .

[0066] Where N represents the total number of RR intervals. Secondly, the Pearson correlation coefficient between adjacent RR intervals can be constructed as follows:

[0067] .

[0068] Where r represents the Pearson correlation coefficient, which characterizes the correlation of variation between adjacent RR intervals.

[0069] In practice, the RR variability of atrial fibrillation (AF) is highly irregular, but arrhythmias (such as premature atrial contractions and premature ventricular contractions) and data noise (such as electrode detachment and electromyographic interference) can also cause RR variability, easily interfering with AF identification. Therefore, we propose an RR interval purification technique based on RF+XGBoost combined classification to ensure high-quality RR variability data. Then, by eliminating non-AF-related interference, the purity of the RR variability data is improved, ensuring high accuracy in subsequent AF identification. Furthermore, traditional deep learning methods (such as CNN and LSTM) are slow in AF detection, processing 24-hour ECG data may take several minutes to tens of minutes, making them unsuitable for practical applications. This method uses a rule-based engine-based feature extraction + RF+XGBoost machine learning model, significantly optimizing computational efficiency, completing AF detection of 24-hour ECG data in just a few seconds. This greatly improves computational speed, far exceeding the computational efficiency of deep learning methods, making it suitable for real-time monitoring and telemedicine. It can also be deployed on ordinary PCs and even embedded devices, suitable for wearable ECG devices, Holter monitoring, hospital ECG equipment, etc.

[0070] The third step is to determine the standard deviation and coefficient of variation for each RR interval in the aforementioned RR interval sequence. The standard deviation characterizes the magnitude of RR variation within the atrial fibrillation interval. The coefficient of variation characterizes the irregularity of RR variation within the atrial fibrillation interval. The coefficient of variation can be generated using the following formula:

[0071] .

[0072] Here, CV represents the coefficient of variation; a high CV value usually indicates disordered RR changes. SDNN represents the standard deviation; a larger value indicates greater variability in heart rhythm. The mean RR interval can be calculated using a 15-beat sliding window.

[0073] Fourth, for each RR interval to be labeled and its corresponding coefficient of variation and standard deviation in the above RR interval sequence, perform the following steps to generate atrial fibrillation event groups:

[0074] 1. In response to the above-mentioned coefficient of variation being greater than a preset coefficient of variation interval and standard deviation being greater than a preset standard deviation interval, the above-mentioned RR interval to be labeled is labeled as a highly variable RR interval. Wherein, a coefficient of variation greater than a preset coefficient of variation interval (e.g., [5%, 12%]) indicates that the RR interval to be labeled is a highly variable RR interval.

[0075] 2. In response to the coefficient of variation being within the preset coefficient of variation range and the standard deviation being within the preset standard deviation range, the RR interval to be labeled is marked as the critical interval for RR variation. Wherein, a coefficient of variation within the preset coefficient of variation range and a standard deviation within the preset standard deviation range (e.g., 50-100 milliseconds) indicate that the RR interval to be labeled is in the critical region of variation.

[0076] 3. In response to the coefficient of variation being less than the preset coefficient of variation interval and the standard deviation being less than the preset standard deviation interval, the RR interval to be labeled is deleted. The fact that the coefficient of variation is less than the preset coefficient of variation interval and the standard deviation is less than the preset standard deviation interval indicates a normal heart rhythm and directly rules out the possibility of atrial fibrillation in the RR interval to be labeled.

[0077] Optionally, performing atrial fibrillation event analysis on the purified RR interval sequences to be labeled to generate atrial fibrillation event sets further includes:

[0078] For each highly variable RR interval, perform the following atrial fibrillation event generation steps:

[0079] The first step is to identify the highly variable RR interval as an atrial fibrillation event in response to the determination that the highly variable RR interval is consistently greater than a preset number of heartbeats. Specifically, a highly variable RR interval consistently greater than a preset number of heartbeats (e.g., 8 heartbeats) is identified as an atrial fibrillation event.

[0080] The second step involves merging atrial fibrillation events corresponding to highly variable RR intervals that are adjacent to the aforementioned highly variable RR intervals and whose intervals are less than a preset duration (e.g., 2 seconds) into a single atrial fibrillation event. Specifically, if the critical RR interval is located between the intervals of two atrial fibrillation events, then the critical RR interval and the two atrial fibrillation events are marked as a continuous atrial fibrillation region.

[0081] As an example, see reference Figure 3 On the purified effective RR interval data (excluding various interferences), a sliding window was used to calculate RR variability and pinpoint the approximate interval of atrial fibrillation.

[0082] In practice, combining event sequence analysis can further optimize the determination of AF start and end points, thereby improving the temporal resolution of AF identification. Furthermore, by using a fixed threshold-based method, it is possible to adapt to different patients' RR variation patterns and avoid misjudgments caused by individual differences.

[0083] Step 105: Perform morphological analysis on the noise-reduced electrocardiogram signal to generate electrocardiogram P-wave features.

[0084] In some embodiments, the execution entity may perform morphological analysis on the noise-reduced electrocardiogram (ECG) signal to generate ECG P-wave features. These ECG P-wave features include the P-wave initiation point, the P-wave peak value, and the P-wave termination point.

[0085] Here, P wave amplitude can be the voltage difference between the P wave peak and the isoelectric line, identifying energy changes in the P wave. P wave duration: Calculate the time interval between the start and end points of the P wave to determine its stability. Shortening or deformation of the P wave duration suggests premature atrial contractions or atrial fibrillation. P wave morphological stability: Calculate the cosine similarity of the P wave morphology to analyze its consistency. Sudden changes or disappearance of the P wave morphology in continuous heartbeats are marked as the start or end point of an atrial fibrillation event. P wave position: Calculate the position of the P wave relative to the R wave (PR interval) to ensure its normal presence. Disappearance of the P wave in multiple heartbeats, coupled with increased RR variability, suggests atrial fibrillation is occurring.

[0086] In some optional implementations of certain embodiments, the execution entity performs morphological analysis on the denoised electrocardiogram signal to generate electrocardiogram P-wave features, including:

[0087] The first step involves using multi-scale wavelet transform to decompose the denoised ECG signal, resulting in a decomposed ECG signal. This decomposition process is used to further remove noise from baseline drift and electromyography (EMG).

[0088] The second step involves enhancing the P-wave features in the decomposed ECG signal to obtain an enhanced ECG signal. This enhancement can be achieved using the db4 (Daubechies) wavelet transform function, making the P-wave features more prominent in the signal.

[0089] The third step involves filtering the enhanced electrocardiogram (ECG) signal to obtain a filtered ECG signal. High-frequency components are used to filter noise, while low-frequency components are used to preserve P-wave morphology information.

[0090] The fourth step is to improve the signal-to-noise ratio (SNR) of the filtered ECG signal to obtain the adjusted ECG signal. This can be achieved by smoothing the filtered ECG signal using a Savitzky-Golay filter to improve the SNR, while maintaining the integrity of the P-wave waveform.

[0091] The fifth step involves extracting P-wave features from the adjusted ECG signal to obtain the P-wave characteristics. Specifically, the voltage difference between the isoelectric lines of the adjusted ECG signal is determined as the P-wave peak. The start and end points of the P-wave feature in the adjusted ECG signal are identified, and the difference between the start and end points is determined as the P-wave duration. The PR interval of the P-wave feature relative to the R wave in the adjusted ECG signal is determined. When the P-wave disappears in multiple heartbeats in the adjusted ECG signal, and the PR interval variability increases, it is marked as the start or end point of an atrial fibrillation event. Here, if the P-wave duration is shortened or deformed, it may indicate the occurrence of premature atrial contractions or atrial fibrillation. The PR interval is the time interval between the start of the P-wave on the ECG and the start of the QRS complex.

[0092] Additionally, the pre-similarity of P wave morphology can be calculated to analyze the consistency of P wave morphology. If the P wave morphology changes abruptly or disappears during continuous heartbeats, it is marked as the beginning or end of an atrial fibrillation event.

[0093] Step 106: Dynamically monitor the P wave characteristics of the electrocardiogram, and adjust the atrial fibrillation onset time and atrial fibrillation termination time of each atrial fibrillation event in the atrial fibrillation event group based on the results of the dynamic monitoring, to obtain the optimized atrial fibrillation signal group.

[0094] In some embodiments, the execution entity can dynamically monitor the electrocardiogram P wave characteristics and, based on the results of the dynamic monitoring, adjust the atrial fibrillation initiation time and atrial fibrillation termination time of each atrial fibrillation event in the atrial fibrillation event group to obtain an optimized atrial fibrillation signal group.

[0095] In some optional implementations of certain embodiments, the execution entity dynamically monitors the electrocardiogram P-wave characteristics and, based on the results of the dynamic monitoring, adjusts the atrial fibrillation onset and termination times of each atrial fibrillation event in the atrial fibrillation event group to obtain an optimized atrial fibrillation signal group, including:

[0096] The first step involves dynamically monitoring the aforementioned ECG P-wave characteristics using a second sliding window to obtain dynamic monitoring results. These results may include: whether the P-wave disappears within the window, the variability of the P-wave waveform within the window, whether the P-wave abnormality is accompanied by an irregular increase in the RR interval, and the abnormal coefficient of the RR interval accompanying the disappearance of the P-wave.

[0097] The second step is to optimize the atrial fibrillation initiation and termination times of each atrial fibrillation event in the atrial fibrillation event group based on the dynamic monitoring results mentioned above, and obtain the optimized atrial fibrillation signal group.

[0098] Specifically, time optimization can be achieved through the following methods:

[0099] If more than 80% of the P waves disappear within the sliding window, it can be preliminarily determined that an AF event may have occurred.

[0100] If the disappearance of the P wave is accompanied by highly irregular changes in the RR interval (i.e., the abnormal coefficient is greater than the preset coefficient of variation range mentioned above), the confidence level of the AF event in this segment can be enhanced.

[0101] If the variability of the P wave exceeds a threshold, it indicates an abnormal P wave. Here, abnormalities may include premature atrial contractions (PACs) or the onset of atrial fibrillation (AF).

[0102] If an abnormal P wave is accompanied by an irregular increase in the RR interval, it can be identified as the early stage of atrial fibrillation (AF), and the AF event confirmation process can then be executed. The AF event confirmation process can either involve re-analyzing the P wave characteristics on the ECG for AF event, or sending the data to a human operator for final review.

[0103] Here, the start and end times of atrial fibrillation events can be optimized through the following steps:

[0104] Calculate the P wave morphological trend of the 10 heartbeats preceding the onset of atrial fibrillation (AF). If the P wave gradually weakens and the CV (volume response) exceeds 10%, it indicates the early stage of AF, and the starting point is adjusted forward. If the P wave suddenly disappears, the heartbeat represents the onset point of AF.

[0105] If, within 5-10 heartbeats after atrial fibrillation is diagnosed, the P wave gradually recovers and the RR interval variability decreases, then this is considered the termination point of atrial fibrillation. If the P wave recovers briefly but the RR interval variability is large, then the atrial fibrillation is persistent or incompletely terminated, and further assessment is needed to determine whether it is paroxysmal atrial fibrillation.

[0106] As an example, see Figure 4 Within the approximate atrial fibrillation region located by the RR interval, beat-by-beat P wave features are extracted, and the atrial fibrillation time point is further precisely identified.

[0107] In practice, traditional deep learning and other conventional methods, based on fixed time windows of 1 minute, 30 seconds, or 10 seconds, cannot pinpoint the specific heartbeat at which atrial fibrillation (AF) occurs. This method employs sliding window RR variability analysis combined with precise P-wave detection to identify AF events at the single-beat level: a 15-beat sliding window is used to calculate the SDNN and CV, filtering for highly variable RR regions. P-wave morphology analysis ensures the accuracy of the AF event's initiation and termination, avoiding errors caused by fixed window divisions. Therefore, it can not only complete AF detection of 24-hour Holter monitoring data within seconds but also achieve accuracy down to the single-beat level. This method breaks through the limitations of fixed time windows, improving the detection accuracy of paroxysmal and persistent AF events. It is suitable for Holter monitoring, wearable devices, and telemedicine applications.

[0108] The various embodiments of this disclosure have the following beneficial effects: the atrial fibrillation signal identification method of some embodiments of this disclosure can improve the accuracy of atrial fibrillation event identification. Specifically, the reason why it is easy to miss or misdiagnose due to arrhythmia or artifact interference is that: traditional manual interpretation methods are greatly affected by subjective factors, the doctor's experience and the level of review of the film directly determine the accuracy of the diagnosis, and for brief AF events lasting less than 1 minute, manual identification is more difficult. Based on this, the atrial fibrillation signal identification method of some embodiments of this disclosure firstly acquires an electrocardiogram (ECG) signal through an ECG acquisition device. Noise is then removed to ensure signal clarity. Secondly, the ECG signal is denoised to generate a denoised ECG signal. Among them, the position of the R-wave peak signal in the denoised ECG signal is marked by R-point detection. The R-point marking can provide basic data for subsequent RR interval calculation. Then, the RR interval in the denoised ECG signal is located to generate a RR interval sequence to be marked. In this study, each RR interval in the aforementioned unlabeled RR interval sequence corresponds to a set of atrial fibrillation onset and termination time points. Preliminary RR interval localization helps determine usable RR intervals and reduces data interference. Subsequently, the unlabeled RR interval sequence is used to identify RR intervals affected by postural changes and regular arrhythmias using XGboost and the Stochastic Forest model. RR interval sequences from these regions are excluded from the identification area, completing the purification process. Atrial fibrillation event analysis is then performed on the purified unlabeled RR interval sequence to generate atrial fibrillation event sets. A sliding window is used to dynamically monitor the variability of each unlabeled RR interval in the purified unlabeled RR interval sequence. This dynamic monitoring helps to preliminarily locate the occurrence intervals of atrial fibrillation. Next, morphological analysis is performed on the noise-reduced electrocardiogram (ECG) signal to generate ECG P-wave features. These ECG P-wave features include the P-wave onset, P-wave peak, and P-wave end. Morphological analysis further refines the P-wave features. Finally, the aforementioned ECG P-wave characteristics were dynamically monitored, and based on the monitoring results, the atrial fibrillation onset and termination times of each atrial fibrillation event in the atrial fibrillation event group were adjusted to obtain an optimized atrial fibrillation signal group. Here, through more precise P-wave characteristics and dynamic monitoring, atrial fibrillation events can be located more accurately, and the corresponding time points can be optimized. This improves the accuracy of atrial fibrillation signal identification.

[0109] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of an atrial fibrillation signal identification device, which are similar to... Figure 1 Corresponding to the method embodiments shown, this atrial fibrillation signal identification device can be specifically applied to various electronic devices.

[0110] like Figure 5 As shown, the atrial fibrillation signal identification device 500 in some embodiments includes: an acquisition unit 501, a noise reduction processing unit 502, a signal positioning unit 503, an atrial fibrillation event analysis unit 504, a morphological analysis unit 505, and a dynamic monitoring and optimization unit 506. The acquisition unit 501 is configured to acquire electrocardiogram (ECG) signals through an ECG acquisition device; the noise reduction processing unit 502 is configured to perform noise reduction processing on the ECG signals to generate noise-reduced ECG signals, wherein the R-wave peak signal position in the noise-reduced ECG signals is marked by R-point detection; the signal localization unit 503 is configured to locate the RR intervals in the noise-reduced ECG signals to generate a sequence of RR intervals to be labeled, wherein each RR interval to be labeled in the sequence of RR intervals to be labeled corresponds to a set of atrial fibrillation initiation time points and atrial fibrillation termination time points; the atrial fibrillation event analysis unit 504 is configured to identify the RR intervals of the sequence of RR intervals to be labeled using XGboost and the Random Forest model to identify RR intervals affected by postural changes and regular arrhythmias, and to analyze these regions. The RR interval sequences are excluded from the identification region to complete the purification process, and atrial fibrillation event analysis is performed on the purified RR interval sequences to generate an atrial fibrillation event group. The variability of each RR interval to be labeled in the purified RR interval sequences is dynamically monitored through a sliding window. The morphological analysis unit 505 is configured to perform morphological analysis on the above-mentioned noise-reduced electrocardiogram signal to generate electrocardiogram P wave features, wherein the above-mentioned electrocardiogram P wave features include P wave onset, P wave peak, and P wave end. The dynamic monitoring and optimization unit 506 is configured to dynamically monitor the above-mentioned electrocardiogram P wave features, and adjust the atrial fibrillation onset time point and atrial fibrillation termination time point of each atrial fibrillation event in the above-mentioned atrial fibrillation event group according to the results of dynamic monitoring, so as to obtain an optimized atrial fibrillation signal group.

[0111] It is understandable that the units described in the atrial fibrillation signal identification device 500 are related to the reference. Figure 1 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the atrial fibrillation signal identification device 500 and the units contained therein, and will not be repeated here.

[0112] The following is for reference. Figure 6 It shows a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of this disclosure. Figure 6As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory may include non-volatile storage media and internal memory. The non-volatile storage media may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any front-end page monitoring method. The processor provides computing and control capabilities to support the operation of the entire computer device. The internal memory provides an environment for the execution of the computer program in the non-volatile storage media; when executed by the processor, this program causes the processor to perform any front-end page monitoring method. The network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the computer device to which the present disclosure is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0113] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0114] In one embodiment, the processor is used to run a computer program stored in a memory to perform the following steps: acquiring an electrocardiogram (ECG) signal through an ECG acquisition device; performing noise reduction processing on the ECG signal to generate a noise-reduced ECG signal, wherein the R-wave peak signal position in the noise-reduced ECG signal is marked by R-point detection; locating the RR intervals in the noise-reduced ECG signal to generate a sequence of RR intervals to be labeled, wherein each RR interval to be labeled in the sequence corresponds to a set of atrial fibrillation initiation time points and atrial fibrillation termination time points; and identifying the RR intervals of the RR interval sequence to be labeled by using XGboost and a randomized deep forest model to identify RR intervals affected by postural changes and regular arrhythmias. The RR interval sequences in these regions are excluded from the identification area to complete the purification process. Atrial fibrillation event analysis is then performed on the purified RR interval sequences to generate an atrial fibrillation event set. The variability of each RR interval in the purified RR interval sequences is dynamically monitored using a sliding window. Morphological analysis is performed on the noise-reduced electrocardiogram (ECG) signals to generate ECG P-wave features, including P-wave onset, P-wave peak, and P-wave end. The ECG P-wave features are then dynamically monitored, and based on the results, the atrial fibrillation onset and termination times of each atrial fibrillation event in the atrial fibrillation event set are adjusted to obtain an optimized atrial fibrillation signal set.

[0115] This disclosure also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can be referred to various embodiments of the atrial fibrillation signal identification method of this disclosure.

[0116] The aforementioned computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. Alternatively, the aforementioned computer-readable storage medium may be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0117] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0118] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for identifying atrial fibrillation signals, comprising: Electrocardiogram (ECG) signals are acquired using ECG acquisition equipment; The electrocardiogram (ECG) signal is denoised to generate a denoised ECG signal, wherein the position of the R-wave peak signal in the denoised ECG signal is marked by R-point detection. The RR intervals in the noise-reduced electrocardiogram signal are located to generate a sequence of RR intervals to be labeled, wherein each RR interval in the sequence of RR intervals to be labeled corresponds to a set of atrial fibrillation initiation time points and atrial fibrillation termination time points; The RR interval sequences to be labeled are identified using XGBoost and a random forest model to remove RR intervals affected by postural changes and regular arrhythmias, thus purifying the sequences. Atrial fibrillation event analysis is then performed on the purified RR interval sequences to generate atrial fibrillation event sets. A sliding window is used to dynamically monitor the variability of each RR interval in the purified sequences. The RR interval sequences are then input into a pre-trained random forest model to obtain preliminary detected RR interval sequences. The random forest model is used to identify the trend of RR interval changes based on the smoothness of the RR interval sequences to determine if there are any sudden abnormal signals. The RR intervals are used to label and remove those that interfere with body position changes and regular arrhythmias. The pre-detected RR interval sequence is input into a pre-trained machine learning model to obtain a sequence of RR intervals to be labeled. The machine learning model is used to remove pre-detected RR intervals that represent electrode detachment or electromyographic interference. The RR intervals to be labeled represent highly variable RR intervals. The standard deviation and coefficient of variation for each RR interval in the sequence to be labeled are determined. The standard deviation represents the magnitude of RR variation in the atrial fibrillation interval, and the coefficient of variation represents the irregularity of RR variation in the atrial fibrillation interval. For each RR interval in the sequence to be labeled and its corresponding coefficient of variation and standard deviation, the following steps are performed to generate atrial fibrillation event groups: In response to the coefficient of variation being greater than a preset coefficient of variation interval and the standard deviation being greater than a preset standard deviation interval, the RR interval to be labeled is marked as a high-variable RR interval. In response to the coefficient of variation being within the preset coefficient of variation range and the standard deviation being within the preset standard deviation range, the RR interval to be labeled is marked as the critical interval of RR variation; In response to the coefficient of variation being less than the preset coefficient of variation interval and the standard deviation being less than the preset standard deviation interval, the RR interval to be labeled is deleted; Morphological analysis is performed on the noise-reduced electrocardiogram signal to generate electrocardiogram P-wave features, wherein the electrocardiogram P-wave features include P-wave initiation, P-wave peak value, and P-wave end value; The P-wave characteristics of the electrocardiogram are dynamically monitored, and the atrial fibrillation onset time and atrial fibrillation termination time of each atrial fibrillation event in the atrial fibrillation event group are adjusted based on the results of the dynamic monitoring to obtain an optimized atrial fibrillation signal group.

2. The method according to claim 1, characterized in that, The step of denoising the electrocardiogram (ECG) signal to generate a denoised ECG signal includes: The electrocardiogram signal was subjected to baseline drift and electromyographic interference removal by bandpass filtering to obtain a noise-reduced electrocardiogram signal. The R-wave peak value of the noise-reduced electrocardiogram (ECG) signal is detected by adaptive threshold detection and second derivative method to mark the position of the R-wave peak signal in the noise-reduced ECG signal.

3. The method according to claim 1, characterized in that, The step of locating the RR intervals in the noise-reduced electrocardiogram signal to generate a sequence of RR intervals to be labeled includes: Obvious abnormalities were removed from the noise-reduced electrocardiogram signal to obtain the RR interval sequence to be labeled.

4. The method according to claim 3, characterized in that, The step of performing atrial fibrillation event analysis on the purified RR interval sequences to be labeled to generate atrial fibrillation event groups also includes: For each highly variable RR interval, perform the following atrial fibrillation event generation steps: In response to determining that the highly variable RR interval is continuously greater than a preset number of heartbeats, the highly variable RR interval is labeled as an atrial fibrillation event; Atrial fibrillation events corresponding to highly variable RR intervals that are adjacent to the highly variable RR interval and whose interval is less than a preset duration are merged into the same atrial fibrillation event. In response to the critical interval of RR variation being between the intervals of two atrial fibrillation events, the critical interval of RR variation and the two atrial fibrillation events are marked as a continuous atrial fibrillation region.

5. The method according to claim 4, characterized in that, The step of performing morphological analysis on the denoised electrocardiogram signal to generate electrocardiogram P-wave features includes: The denoised electrocardiogram signal is decomposed using multi-scale wavelet transform to obtain the decomposed electrocardiogram signal, wherein the decomposition is used to remove noise from baseline drift and electromyography for a second time; The P-wave features in the decomposed electrocardiogram signal are enhanced to obtain an enhanced electrocardiogram signal. The enhanced electrocardiogram signal is filtered to obtain a filtered electrocardiogram signal, wherein the high-frequency component is used to filter noise and the low-frequency component is used to preserve P wave morphology information. The signal-to-noise ratio of the filtered electrocardiogram signal is increased to obtain the adjusted electrocardiogram signal; The P-wave features are extracted from the adjusted electrocardiogram (ECG) signal to obtain the ECG P-wave features. Specifically, the voltage difference between the isoelectric lines of the adjusted ECG signal is determined as the P-wave peak, which is also determined as the P-wave start and end point of the P-wave features in the adjusted ECG signal. The difference between the P-wave start and end point is determined as the P-wave duration. The PR interval of the P-wave features relative to the R wave in the adjusted ECG signal is determined. When the P-waves of multiple heartbeats disappear in the adjusted ECG signal and the PR interval variability increases, it is marked as the start or end point of an atrial fibrillation event.

6. The method according to claim 5, characterized in that, The process of dynamically monitoring the P-wave characteristics of the electrocardiogram and adjusting the atrial fibrillation onset and termination times of each atrial fibrillation event in the atrial fibrillation event group based on the results of the dynamic monitoring to obtain an optimized atrial fibrillation signal group includes: The P-wave characteristics of the electrocardiogram are dynamically monitored through a second sliding window to obtain dynamic monitoring results. Based on the dynamic monitoring results, the atrial fibrillation onset time and atrial fibrillation termination time of each atrial fibrillation event in the atrial fibrillation event group were optimized to obtain the optimized atrial fibrillation signal group.

7. An atrial fibrillation signal identification device, comprising: The acquisition unit is configured to acquire electrocardiogram (ECG) signals via an ECG acquisition device; The noise reduction processing unit is configured to perform noise reduction processing on the electrocardiogram signal to generate a noise-reduced electrocardiogram signal, wherein the position of the R-wave peak signal in the noise-reduced electrocardiogram signal is marked by R-point detection. A signal localization unit is configured to locate the RR intervals in the denoised electrocardiogram signal to generate a sequence of RR intervals to be labeled, wherein each RR interval in the sequence corresponds to a set of atrial fibrillation onset and termination time points; an atrial fibrillation event analysis unit is configured to use XGboost and a random forest model to identify RR intervals affected by postural changes and regular arrhythmias in the sequence of RR intervals to be labeled, exclude RR interval sequences in these regions from the identification area to complete the purification process, and perform atrial fibrillation event analysis on the purified sequence of RR intervals to be labeled to generate atrial fibrillation event groups, wherein the variability of each RR interval in the purified sequence of RR intervals to be labeled is dynamically monitored through a sliding window, and the sequence of RR intervals to be labeled is input into a pre-trained random forest model to obtain a preliminary detected RR interval sequence. The random forest model is used to identify the trend of RR interval changes based on the smoothness of the RR interval sequence to determine whether there are sudden abnormal signals, so as to label and remove RR intervals that interfere with body position changes and regular arrhythmias; the RR interval sequence after preliminary detection is input into a pre-trained machine learning model to obtain the RR interval sequence to be labeled, wherein the machine learning model is used to remove the RR intervals after preliminary detection that represent electrode detachment or electromyographic interference, and the RR intervals to be labeled represent RR intervals with high variability; the standard deviation and coefficient of variation corresponding to each RR interval in the RR interval sequence to be labeled are determined, wherein the standard deviation is used to represent the magnitude of RR variation in the atrial fibrillation interval, and the coefficient of variation is used to represent the irregularity of RR variation in the atrial fibrillation interval; for each RR interval to be labeled and the corresponding coefficient of variation and standard deviation in the RR interval sequence to be labeled, the following steps are performed to generate atrial fibrillation event groups: In response to the coefficient of variation being greater than a preset coefficient of variation interval and the standard deviation being greater than a preset standard deviation interval, the RR interval to be labeled is marked as a high-variable RR interval. In response to the coefficient of variation being within the preset coefficient of variation range and the standard deviation being within the preset standard deviation range, the RR interval to be labeled is marked as the critical interval of RR variation; In response to the coefficient of variation being less than the preset coefficient of variation interval and the standard deviation being less than the preset standard deviation interval, the RR interval to be labeled is deleted; A morphological analysis unit is configured to perform morphological analysis on the denoised electrocardiogram signal to generate electrocardiogram P-wave features, wherein the electrocardiogram P-wave features include P-wave initiation, P-wave peak value, and P-wave end value. The dynamic monitoring and optimization unit is configured to dynamically monitor the P-wave characteristics of the electrocardiogram and, based on the results of the dynamic monitoring, adjust the atrial fibrillation initiation time and atrial fibrillation termination time of each atrial fibrillation event in the atrial fibrillation event group to obtain an optimized atrial fibrillation signal group.

8. An electronic device, comprising: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

9. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images