Research on Identifying and Avoiding Overdiagnosis of ADHD in Students

Attention Deficit Hyperactivity Disorder (ADHD) is a common neurodevelopmental disorder that affects the physical and mental development. The traditional recognition method relying on scale evaluation is easily influenced by subjective factors, and there is a risk of misjudgment and overdiagnosis. This paper proposes a student ADHD assisted recognition method based on the fusion of eye movement and facial expression temporal features. This article uses the eye tracking MPIIFaceGaze dataset and facial CK+dataset to construct an EFF-ARM recognition model using Transformer network. The experiment adopts a cross student testing method and compares it with CNN and RNN models for analysis. The results showed that the accuracy, precision, and recall of the EFF-ARM model reached 86.42%, 85.76%, and 85.91%, respectively, and the overall recognition performance was better than the comparison model. Finally, this article implemented a system platform for identifying and avoiding overdiagnosis of ADHD in students, which achieved feature analysis and risk result display, providing an auxiliary analysis method based on objective behavioral data for early screening of ADHD in students.