Early screening is vital and helpful for implementing intensive intervention and rehabilitation therapy for children with autism spectrum disorder (ASD). Research has shown that electroencephalogram (EEG) signals can reflect abnormal brain function of children with ASD, and screening with EEG signals has the characteristics of good real-time performance and high sensitivity. However, the existing EEG screening algorithms mostly focus on the data analysis in the resting state, and the extracted EEG features have some disadvantages such as weak representation capacity and information redundancy. In this study, we utilized the event-related potential (ERP) technique to acquire the EEG data of the subjects under positive and negative emotional stimulation and proposed an EEG Feature Selection Algorithm based on L1-norm regularization to perform screening of autism. The proposed EEG Feature Selection Algorithm includes the following steps: (1) extracting 20 EEG features from the raw data, (2) classification with support vector machine, (3) selecting appropriate EEG feature with L1-norm regularization according to the classification performance. The experimental results show that the accuracy for screening of children with ASD can reach 93.8% and 87.5% under positive and negative emotional stimulation and the proposed algorithm can effectively eliminate redundant features and improve screening accuracy.

Background: Facial behavior has emerged as a crucial biomarker for autism identification. However, heterogeneity among individuals with autism poses a significant obstacle to traditional feature extraction methods, which often lack the necessary discriminative power. While deep-learning methods hold promise, they are often criticized for their lack of interpretability. Methods: To address these challenges, we developed an innovative facial behavior characterization model that integrates coarse- and fine-grained analyses for intelligent autism identification. The coarse-grained analysis provides a holistic view by computing statistical measures related to facial behavior characteristics. In contrast, the fine-grained component uncovers subtle temporal fluctuations by employing a long short-term memory (LSTM) model to capture the temporal dynamics of head pose, facial expression intensity, and expression types. To fully harness the strengths of both analyses, we implemented a feature-level attention mechanism. This not only enhances the model’s interpretability but also provides valuable insights by highlighting the most influential features through attention weights. Results: Upon evaluation using three-fold cross-validation on a self-constructed autism dataset, our integrated approach achieved an average recognition accuracy of 88.74%, surpassing the standalone coarse-grained analysis by 8.49%. Conclusions: This experimental result underscores the improved generalizability of facial behavior features and effectively mitigates the complexities stemming from the pronounced intragroup variability of those with autism, thereby contributing to more accurate and interpretable autism identification.