A study on promoting AI learning and usage behaviors among health management students from the perspective of the “knowledge-belief-action” model

Introduction: The application of Artificial Intelligence (AI) in the field of health management is becoming increasingly widespread, driving health services toward precision, personalization, and intelligence. This study aims to explore the behavior patterns of health management college students in AI learning and device usage. In addition, the study also aims to enhance students' motivation and ability to apply AI by constructing a feedback oriented behavior promotion system based on machine learning technology. Methods: This study targeted college students from the School of Health Management at Zaozhuang University. A survey questionnaire based on the "Knowledge-Belief-Action" (KBA) model was designed, and a total of 184 valid responses were collected. The XGBoost algorithm was used to construct a predictive model for AI learning and usage behavior, and SHAP technology was applied for interpretative analysis of the model results to identify key influencing factors. Furthermore, the model was integrated into a web platform, and a visualized behavior promotion system was developed. Results: The accuracy, precision, recall, and F1-score of the predictive model all exceeded 0.698, indicating strong predictive capability. SHAP analysis revealed that factors such as knowledge mastery, awareness of ethical issues, and educational background have a significant impact on students' AI learning and usage behavior. The behavior promotion system developed based on this model not only predicts students' learning and usage behaviors but also provides a basis for personalized intervention. Discussion: This study combines the "KBA" model with machine learning to construct an interpretable predictive model for students' AI learning and usage behavior. The study shows that ethical awareness, educational background, and practical application experience are important factors influencing students' behavior. Based on this model, we further developed a behavior promotion system, providing new ideas and tools for optimizing AI education in universities. However, this study also has limitations, such as a single source of sample data. Future research could expand the sample range to further verify the generalizability of the research conclusions.