Research on student behavior recognition method based on human physiological information perception

Artificial intelligence technology can assist teaching and learning by monitoring students’ learning behavior, while the traditional methods of recognizing student behavior suffer from environmental limitations, data singularity, and subjective intervention by the experimenter. Therefore, this paper proposed a student behavior recognition method based on the perception of human physiological information. The method first gathers physiological information about students’ body temperature, heart rate, and blood pressure using the smart bracelet and constructs a student behavior dataset based on the three behaviors of students’ study, entertainment, and sleep for behavioral annotation and preprocessing of the data. Second, three methods of feature extraction are applied: the K-means algorithm, the Apriori algorithm, and statistical features. Finally, using a random forest classifier, a student behavior recognition model based on physiological data is built for experimental comparison and evaluation. The experimental results show that under a single feature extraction method, the feature extraction using the Apriori algorithm improves the model performance most significantly, the Random Forest classifier is the most effective for student behavior recognition, and the model accuracy reaches 96.72% after using the three feature extraction methods at the same time. Furthermore, this paper builds a system for recognizing student behavior based on human physiological data and validates the model’s efficacy and real-time performance in authentic scenarios. The proposed method allows educators to better assist students’ learning and manage them holistically.