Electrocardiogram-signal-based assisted identification method, apparatus and system for attention deficit hyperactivity disorder
By combining deep learning and various machine learning methods, deep features of electrocardiogram signals are extracted and interpretable heatmaps are generated, which solves the problems of subjectivity and high cost in ADHD diagnosis and achieves high-precision, low-cost ADHD screening and diagnosis.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- ZHEJIANG UNIV
- Filing Date
- 2025-11-11
- Publication Date
- 2026-06-18
AI Technical Summary
Existing technologies for ADHD diagnosis suffer from high subjectivity, time-consuming nature, and high cost, making it difficult to meet the needs of large-scale screening and early intervention. Furthermore, existing methods struggle to capture the complex nonlinear characteristics of ECG signals, resulting in limited classification accuracy and insufficient robustness.
By combining deep learning with various machine learning methods, deep features of electrocardiogram signals are extracted through a one-dimensional convolutional neural network, and a classification heatmap is generated using Score-CAM. Time domain, frequency domain, and local statistical features are extracted from the heatmap and input into a machine learning classifier for ADHD risk assessment.
It improves the classification accuracy of ADHD diagnosis, enhances the interpretability and robustness of the model, reduces costs, is suitable for clinical outpatient screening, and achieves rapid and objective ADHD screening and diagnosis.
Smart Images
Figure CN2025134009_18062026_PF_FP_ABST
Abstract
Description
Methods, devices, and systems for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals. Technical Field
[0001] This invention relates to the field of medical technology, and in particular to a method, apparatus and system for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals. Background Technology
[0002] Attention deficit hyperactivity disorder (ADHD) is a common neurodevelopmental disorder in children, characterized primarily by inattention, impulsivity, and hyperactivity. ADHD affects approximately 5% to 7% of children and adolescents, and its causes involve a variety of factors, including genetics, neurophysiology, and environment. This disorder not only significantly impacts patients' daily lives, learning, and social interactions but can also persist into adulthood, leading to reduced work efficiency and impaired social functioning. Furthermore, ADHD is often accompanied by comorbidities such as anxiety disorders, conduct disorder, and oppositional defiant disorder, further exacerbating the psychological and physical burden on patients.
[0003] Currently, the diagnosis of ADHD mainly relies on clinical questionnaires (such as the DSM-5 assessment criteria) and behavioral observation. However, these methods are highly subjective, and diagnostic results are easily affected by the assessor's experience and fluctuations in patient performance. Furthermore, traditional methods are time-consuming and costly, making it difficult to meet the clinical needs for large-scale screening and early intervention. Especially in areas with limited medical resources, diagnostic delays or misdiagnoses are frequent. Therefore, there is an urgent need to develop an efficient and objective auxiliary diagnostic tool to improve diagnostic accuracy and shorten diagnosis time.
[0004] In recent years, electrocardiogram (ECG) signals have become a hot topic in the study of the physiological characteristics of ADHD due to their close correlation with autonomic nervous system activity. Existing research has shown that ECG signals from ADHD patients exhibit significant abnormalities, such as a general decrease in heart rate variability (HRV) and irregular signal rhythms. However, most existing studies are based on linear time-domain and frequency-domain analysis methods, which struggle to capture the complex nonlinear characteristics of the signals. Furthermore, these methods have limited classification accuracy and lack robustness when dealing with large-scale datasets.
[0005] Furthermore, while machine learning and deep learning technologies have seen increasingly deeper applications in medical diagnosis in recent years, their application in ADHD research remains in its early stages. For example, some studies have attempted to classify ADHD using simple statistical features combined with traditional machine learning algorithms, but insufficient feature extraction has limited classification performance. For instance, Chinese patent document CN118983076A discloses an auxiliary diagnostic method and system for childhood attention deficit hyperactivity disorder (ADHD). This method includes the following steps: setting up cognitive ability tests based on the cognitive characteristics and developmental status of the children; collecting cognitive ability test data from subjects in the ADHD diagnostic group and the normal control group; preprocessing the collected cognitive ability test data; constructing a support vector machine (SVM) model; and using the trained SVM model to implement the auxiliary diagnosis of ADHD in children. On the other hand, although deep learning technology can extract deep-level features of signals, its "black box" problem and insufficient model interpretability also limit its widespread clinical application.
[0006] Therefore, combining deep learning and machine learning technologies to improve model classification performance while enhancing its clinical interpretability has become a significant technical challenge in current ADHD research. Summary of the Invention
[0007] This invention provides a device for identifying attention deficit hyperactivity disorder based on electrocardiogram signals. It combines deep learning with various machine learning methods, which not only improves classification performance but also provides rich feature interpretability.
[0008] The technical solution of the present invention is as follows:
[0009] A method for identifying attention deficit hyperactivity disorder (ADHD) based on electrocardiogram (ECG) signals includes:
[0010] (1) Collect and process electrocardiogram data of subjects who need to be assessed for the risk of attention deficit hyperactivity disorder;
[0011] (2) A one-dimensional convolutional neural network was used to extract deep features from the processed electrocardiogram data of the test subjects; a classification heatmap was generated from the feature map of the convolutional neural network using Score-CAM, and time domain, frequency domain and local statistical features were extracted from the generated classification heatmap;
[0012] (3) Input the time domain, frequency domain and local statistical features into the machine learning classifier for classification to obtain the assessment results of the risk of the test subject having attention deficit hyperactivity disorder.
[0013] In step (1), the processing of electrocardiogram data includes: taking a screenshot of the signal segment from the 3rd to the 9th second of lead II in the electrocardiogram data as the processed electrocardiogram data.
[0014] In step (2), the one-dimensional convolutional neural network includes four convolutional layers and a fully connected layer connected in sequence. Each convolutional layer is followed by a batch normalization (BatchNorm) and ReLU activation function and a max pooling layer.
[0015] The first convolutional layer uses a convolutional kernel with 3 filters and a kernel size of 27 to capture the basic local features of the ECG signal;
[0016] The second convolutional layer uses a convolutional kernel with 10 filters and a kernel size of 15 to extract the temporal correlation features of the ECG signal.
[0017] The third convolutional layer uses a convolutional kernel with 10 filters and a kernel size of 3 to further extract the mid-order features of the signal;
[0018] The fourth convolutional layer uses a convolutional kernel with 10 filters and a kernel size of 3, and is responsible for capturing the high-order nonlinear features of the ECG signal;
[0019] Each convolutional layer is followed by batch normalization (BatchNorm) and ReLU activation functions, and the feature dimension is reduced through max pooling layers. Finally, the features are flattened through fully connected layers to form deep features.
[0020] The training process of a one-dimensional convolutional neural network includes: collecting and processing electrocardiogram data from several healthy control children (HC) and children with attention deficit hyperactivity disorder (ADHD) to construct a training dataset; and using the training dataset to train the one-dimensional convolutional neural network.
[0021] The training dataset was constructed by collecting 12-lead ECG signal data from healthy control children aged 6 to 12 years and children with attention deficit hyperactivity disorder from real clinical settings, excluding drug intervention and other disease interference factors; and extracting signal segments from lead II for a specific time period (from the 3rd second to the 9th second) to construct the training dataset.
[0022] Furthermore, the one-dimensional convolutional neural network was trained using the Adam optimizer with a learning rate of 0.0002.
[0023] Furthermore, L2 regularization and Dropout techniques are combined during the training of one-dimensional convolutional neural networks to improve the generalization performance of the model.
[0024] In step (2), Score-CAM generates a classification heatmap by weighting the activation maps of each convolutional layer of the one-dimensional convolutional neural network.
[0025] The time-domain features include the global activation mean, the standard deviation of activation values, and the maximum activation value.
[0026] The global activation mean reflects the average level of activation values across the entire heatmap, used to measure the overall activity of the signal. The activation mean in the ADHD group is typically higher than that in the healthy control group, indicating greater signal fluctuations. The standard deviation of activation values represents the distribution range of activation values on the heatmap; the standard deviation in the ADHD group is significantly higher than that in the HC group, revealing the instability of its signal characteristics. The maximum activation value, the location of the strongest activation value on the heatmap, reflects the region of greatest interest to the model, typically concentrated near the QRS complex and T wave.
[0027] The frequency domain features include the Fourier transform mean, peak spectral value, and peak power.
[0028] The Fourier transform mean is used to analyze the spectral distribution of the signal corresponding to the heatmap. The ADHD group typically exhibits a higher spectral mean, indicating the complexity of the signal's frequency components. Spectral peaks capture the most prominent frequency components of the signal; the ADHD group shows greater fluctuations in this characteristic compared to the HC group. Peak power represents the concentration of signal energy in the frequency domain; the ADHD group has higher peak power than the HC group, reflecting more pronounced energy fluctuations in its signal.
[0029] Local statistical characteristics include sliding window mean, sliding window standard deviation, and local maximum.
[0030] The sliding window mean is the average level of local activation values calculated using a sliding window. The sliding mean of the ADHD group fluctuates more, reflecting the strong changes in the signal over different time periods. The sliding window standard deviation quantifies the volatility of local activation values; it is significantly higher in the ADHD group than in the HC group, indicating the instability of their signal on small time scales. The local maxima are the largest activation values within the sliding window, identifying key regions of the local signal; the ADHD group typically has higher local maxima.
[0031] Statistical analysis of the heatmap characteristics revealed significant differences in signal properties between the ADHD group and the HC group: Time-domain analysis showed that the ADHD group exhibited higher overall activation levels and volatility, reflecting the disordered characteristics of the patients' autonomic nervous system activity. Frequency-domain analysis revealed the diversity and complexity of the spectral components in the ADHD group, indicating stronger signal irregularity. Local statistical characteristics further reflected that the dynamic changes in the electrocardiogram signals of ADHD patients were more pronounced in the time dimension.
[0032] In step (3), the machine learning classifier is a support vector machine (SVM), random forest (RF), logistic regression (LR), k-nearest neighbors (KNN), decision tree (DT), or XGBoost.
[0033] Training a machine learning classifier includes:
[0034] Electrocardiogram (ECG) data were collected from several healthy control children and children with attention deficit hyperactivity disorder (ADHD) and processed. The processed ECG data were then divided into a training set and a test set.
[0035] The training set is input into a trained one-dimensional convolutional neural network. Score-CAM is used to generate a classification heatmap from the feature map of the convolutional neural network. Time domain, frequency domain and local statistical features are extracted from the generated classification heatmap.
[0036] The extracted time-domain, frequency-domain, and local statistical features are input into the machine learning classifier to train the machine learning classifier.
[0037] The present invention also provides a device for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals, comprising:
[0038] The data acquisition and processing module collects and processes electrocardiogram data from subjects whose risk of attention deficit hyperactivity disorder needs to be assessed.
[0039] The heatmap feature extraction module uses a one-dimensional convolutional neural network to extract deep features from the processed electrocardiogram data of the test subjects; it generates a classification heatmap from the feature map of the convolutional neural network through Score-CAM, and extracts time domain, frequency domain and local statistical features from the generated classification heatmap.
[0040] The classification and judgment module inputs time domain, frequency domain, and local statistical features into a machine learning classifier for classification, and obtains the assessment results of the risk of the test subject having attention deficit hyperactivity disorder.
[0041] The present invention also provides a system for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals.
[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0043] (1) High accuracy: Experimental results show that the method of the present invention achieves a classification accuracy of over 95% on clinical datasets.
[0044] (2) Interpretability: The heat map generated by Score-CAM technology improves the interpretability of the model, helps doctors understand the decision-making basis of the model, and makes the diagnostic results of the model more transparent and credible.
[0045] (3) Universality: Multiple machine learning classifiers are used to verify the extracted features, which improves the robustness and reliability of the model.
[0046] (4) Low cost and non-invasive: Compared with traditional imaging and behavioral assessment methods, ECG signal-based methods have the advantages of low cost and ease of use in clinical outpatient screening. Attached Figure Description
[0047] Figure 1 is a schematic diagram of the process of the present invention for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals;
[0048] Figure 2 is a schematic diagram of the 1D-CNN model structure;
[0049] Figure 3 shows the histograms of the heatmap characteristics of the ADHD and HC groups; (a) is the histogram of the average activation value distribution; (b) is the histogram of the maximum activation value distribution; (c) is the histogram of the standard deviation of activation values distribution; (d) is the histogram of the Fourier transform mean distribution; (e) is the histogram of the Fourier transform standard deviation distribution; (f) is the histogram of the maximum Fourier transform value distribution; (g) is the histogram of the power mean distribution; (h) is the histogram of the power maximum distribution; (i) is the histogram of the local mean distribution; (j) is the histogram of the local standard deviation distribution; (k) is the histogram of the local maximum value distribution.
[0050] Figure 4 shows a comparison of the cross-validation performance of 1D-CNN and classifiers; (a) is a line graph comparing the accuracy of 10-fold cross-validation; (b) is a line graph comparing the precision of 10-fold cross-validation; (c) is a line graph comparing the recall of 10-fold cross-validation; (d) is a line graph comparing the F1 score of 10-fold cross-validation.
[0051] Figure 5 shows examples of 1D-Score-CAM heatmaps for ADHD samples (A) and HC samples (B). Detailed Implementation
[0052] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be noted that the embodiments described below are intended to facilitate the understanding of the present invention and do not limit it in any way.
[0053] This invention employs deep analysis of children's electrocardiogram (ECG) signals, combined with a one-dimensional convolutional neural network (1D-CNN) and various machine learning classifiers, to design an automatic detection technology for attention deficit hyperactivity disorder (ADHD) based on ECG signals (flowchart shown in Figure 1). This technology uses heatmaps generated by Score-CAM as its core, constructing a feature extraction method that combines time-domain, frequency-domain, and local statistical features. The classifier accurately classifies ADHD cases from those of healthy individuals based on their ECG characteristics. By combining 1D-CNN deep feature extraction with interpretability techniques, this invention significantly improves the transparency and clinical applicability of the model's classification performance, ultimately forming an efficient and interpretable ADHD detection workflow. This allows clinicians to achieve rapid screening and preliminary diagnosis without relying on behavioral assessments.
[0054] The specific technical solutions of the present invention are further illustrated below through examples.
[0055] Devices that assist in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals include:
[0056] (1) ECG data acquisition and processing module
[0057] In this embodiment, ECG data of children aged 6 to 12 years without drug intervention were collected from a real clinical setting. The total data included 2,368 children diagnosed with ADHD and 2,368 healthy control children (HC). All participants were matched 1:1 by age, and patients with cardiovascular diseases that might affect ECG characteristics were excluded to ensure the authenticity and representativeness of the data.
[0058] The acquired ECG signals were recorded using 12 leads. To improve signal quality and consistency, only lead II signals were used as research data, with a sampling frequency of 300Hz. The signal segment from the 3rd to the 9th second was extracted to form an analysis sample of 3000 sampling points per segment. The signals were not filtered or denoised, preserving all feature information from the original data.
[0059] (2) Deep learning feature extraction module
[0060] A one-dimensional convolutional neural network (1D-CNN) was constructed, and signal features were extracted step by step through four convolutional and pooling layers (the model structure is shown in Figure 2). The specific process is as follows:
[0061] The first convolutional layer uses a 3-filter convolutional kernel with a kernel size of 27 to capture the basic local features of the ECG signal. Batch normalization and ReLU activation function are used to enhance the model's stability and nonlinear expressive power.
[0062] The second convolutional layer uses a kernel with 10 filters and a kernel size of 15 to extract temporal correlation features of the ECG signal. The pooling layer uses 2×2 max pooling to reduce feature dimensionality and retain key information.
[0063] The third convolutional layer uses a convolutional kernel with 10 filters and a kernel size of 3 to further extract the mid-order patterns of the signal and reduce the dimensionality through pooling operations.
[0064] The fourth convolutional layer uses a kernel with 10 filters and a kernel size of 3, responsible for capturing high-order nonlinear features of the ECG signal. It combines batch normalization and pooling operations to improve the compactness and expressiveness of the features.
[0065] Each convolutional layer is followed by batch normalization (BatchNorm) and ReLU activation, and max pooling layers are used to reduce feature dimensionality and mitigate overfitting. Finally, the features are flattened through fully connected layers to form deep features.
[0066] The training of the one-dimensional convolutional neural network uses the Adam optimizer with a learning rate of 0.0002. During training, L2 regularization and Dropout techniques are combined to improve the generalization performance of the model.
[0067] (3) Heatmap feature extraction and interpretability module
[0068] To enhance the clinical interpretability of the model, this invention employs Score-CAM technology to generate classification heatmaps from the feature maps of convolutional neural networks. Score-CAM calculates the contribution of each feature region to the classification result, visually presenting the decision-making process of the deep learning model in heatmap form, thus aiding in understanding the signal regions the model focuses on.
[0069] The time domain, frequency domain, and local statistical features (feature histogram shown in Figure 3) are extracted from the generated heatmap, as follows:
[0070] Time-domain characteristics: 1) Global activation mean: Reflects the average level of the overall activation values in the heatmap, used to measure the overall activity of the signal. The activation mean of the ADHD group is usually higher than that of the healthy control group (HC), indicating that its signal fluctuates more. 2) Standard deviation of activation values: Represents the distribution range of activation values in the heatmap. The standard deviation of the ADHD group is significantly higher than that of the HC group, revealing the instability of its signal characteristics. 3) Maximum activation value: The location of the strongest activation value in the heatmap reflects the region of greatest interest to the model, usually concentrated near the QRS complex and T wave.
[0071] Frequency domain characteristics: 1) Fourier transform mean: Used to analyze the spectral distribution of the signal corresponding to the heatmap. The ADHD group usually exhibits a higher spectral mean, indicating the complexity of the signal's frequency components. 2) Spectral peak value: Captures the most significant frequency components in the signal. The ADHD group shows greater fluctuations in this characteristic compared to the HC group. 3) Power peak value: Indicates the degree of concentration of signal energy in the frequency domain. The power peak value of the ADHD group is higher than that of the HC group, reflecting more severe energy fluctuations in its signal.
[0072] Local statistical characteristics: 1) Sliding window mean: Calculated using a sliding window to measure the average level of local activation values. The sliding mean of the ADHD group fluctuates more, reflecting strong signal changes over different time periods. 2) Sliding window standard deviation: Used to quantify the volatility of local activation values. The ADHD group has a significantly higher standard deviation than the HC group, indicating instability of its signal on smaller time scales. 3) Local maxima: The maximum activation value within the sliding window identifies key regions of the local signal. The ADHD group typically has higher local maxima.
[0073] Statistical analysis of the heatmap characteristics revealed significant differences in signal properties between the ADHD group and the HC group: Time-domain analysis showed that the ADHD group exhibited higher overall activation levels and volatility, reflecting the disordered characteristics of the patients' autonomic nervous system activity. Frequency-domain analysis revealed the diversity and complexity of the spectral components in the ADHD group, indicating stronger signal irregularity. Local statistical characteristics further reflected that the dynamic changes in the electrocardiogram signals of ADHD patients were more pronounced in the time dimension.
[0074] The statistical results of heatmap features significantly improved the accuracy and robustness of ADHD classification, laying a solid foundation for the clinical application of the model.
[0075] (4) ADHD Classification and Performance Verification Module
[0076] The extracted heatmap features were input into the following six machine learning classifiers for classifying ADHD and HC: Support Vector Machine (SVM); Random Forest (RF); Logistic Regression (LR); K-Nearest Neighbors (KNN); Decision Tree (DT); and XGBoost. The classifier performance was evaluated using 10-fold cross-validation, which involved splitting the dataset and repeatedly training and testing to comprehensively assess the model's stability and robustness.
[0077] The results show that the combination of 1D-CNN and random forest classifier performs best (the performance differences of the classifiers are shown in Figure 4), with the following performance metrics: classification accuracy: 95.97%; precision: 0.96; recall: 0.96; F1 score: 0.96.
[0078] To further verify the reliability of the classification results and the interpretability of the model, samples from one ADHD subject and one HC subject were randomly selected for heatmap visualization analysis (sample ECG heatmaps for ADHD and HC are shown in Figure 5). The results showed that in the ADHD group, the activation patterns on the heatmap were more dispersed, with the model primarily focusing on the QRS complex and T wave regions. This indicates that the ECG signals of ADHD patients exhibit significant abnormal fluctuations in these regions. In the HC group, activation values were concentrated near the R wave, and the signal distribution was relatively stable, reflecting the regularity and consistency of ECG signals in healthy individuals.
[0079] The embodiments described above provide a detailed explanation of the technical solutions and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals, characterized in that, include: (1) Collect and process electrocardiogram data of subjects who need to be assessed for the risk of attention deficit hyperactivity disorder; (2) A one-dimensional convolutional neural network was used to extract deep features from the processed electrocardiogram data of the test subjects; Score-CAM was used to weight the activation maps of each convolutional layer of the one-dimensional convolutional neural network to generate a classification heatmap, and time domain, frequency domain and local statistical features were extracted from the generated classification heatmap; The time-domain features include global activation mean, activation value standard deviation, and maximum activation value; The frequency domain characteristics include Fourier transform mean, peak spectral density, and peak power. Local statistical characteristics include sliding window mean, sliding window standard deviation, and local maximum. (3) Input the time domain, frequency domain and local statistical features into the machine learning classifier for classification to obtain the assessment results of the risk of the test subject having attention deficit hyperactivity disorder; The machine learning classifiers mentioned are support vector machines, random forests, logistic regression, K-nearest neighbors, decision trees, or XGBoost.
2. The method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals according to claim 1, characterized in that, In step (1), the processing of electrocardiogram data includes: taking a screenshot of the signal segment from the 3rd to the 9th second of lead II in the electrocardiogram data as the processed electrocardiogram data.
3. The method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals according to claim 1, characterized in that, In step (2), the one-dimensional convolutional neural network includes four convolutional layers and a fully connected layer connected in sequence. Each convolutional layer is followed by a batch normalization and ReLU activation function and a max pooling layer.
4. The method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals according to claim 1, characterized in that, The training process of a one-dimensional convolutional neural network includes: collecting and processing electrocardiogram data from several healthy control children and children with attention deficit hyperactivity disorder to construct a training dataset; and using the training dataset to train the one-dimensional convolutional neural network.
5. The method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals according to claim 1, characterized in that, Training a machine learning classifier includes: Electrocardiogram (ECG) data were collected from several healthy control children and children with attention deficit hyperactivity disorder (ADHD) and processed. The processed ECG data were then divided into a training set and a test set. The training set is input into a trained one-dimensional convolutional neural network. Score-CAM is used to generate a classification heatmap from the feature map of the convolutional neural network. Time domain, frequency domain and local statistical features are extracted from the generated classification heatmap. The extracted time-domain, frequency-domain, and local statistical features are input into the machine learning classifier to train the machine learning classifier.
6. A device for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals, comprising: The data acquisition and processing module collects and processes electrocardiogram data from subjects whose risk of attention deficit hyperactivity disorder needs to be assessed. The heatmap feature extraction module uses a one-dimensional convolutional neural network to extract deep features from the processed electrocardiogram data of the test subjects. It then uses Score-CAM to weight the activation maps of each convolutional layer of the one-dimensional convolutional neural network to generate a classification heatmap. From the generated classification heatmap, it extracts time-domain, frequency-domain, and local statistical features. The time-domain features include the global activation mean, activation standard deviation, and maximum activation value. The frequency-domain features include the Fourier transform mean, peak spectral value, and peak power value. The local statistical features include the sliding window mean, sliding window standard deviation, and local maximum value. The classification and judgment module inputs time domain, frequency domain, and local statistical features into a machine learning classifier for classification to obtain an assessment result of the risk of the test subject having attention deficit hyperactivity disorder; the machine learning classifier is a support vector machine, random forest, logistic regression, K-nearest neighbor algorithm, decision tree, or XGBoost.
7. A system for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for assisting in the identification of attention deficit hyperactivity disorder based on electrocardiogram signals as described in any one of claims 1-5.