THE SOCIOLOGICAL AND HEALTHCARE INTERVENTIONS FOR EXERCISE ADDICTION USING GENERATIVE ARTIFICIAL INTELLIGENCE

This work aims to design and validate an innovative hybrid framework that combines the Support Vector Machine (SVM) for the precise identification of exercise addiction with personalized interventions based on Reinforcement Learning (RL). This approach effectively addresses the growing issue of exercise addiction and fosters the development of healthy exercise habits in individuals. 100 participants are recruited, and high-quality data, including physical activity levels and heart rate, are obtained through collaboration with a fitness tracker manufacturer. Additionally, a self-report questionnaire is designed to gather information on participants’ exercise habits and psychological states, and medical records are integrated to enhance the data comprehensiveness. The SVM model is adopted to classify and identify exercise addiction behaviors, achieving high-precision individual state assessments. Later, based on the identification results, this work develops an intelligent recommendation system based on RL to dynamically adjust and personalize intervention strategies. The results reveal that the intervention led to a significant 35.9% reduction in weekly excessive exercise hours, demonstrating the effectiveness of the measures in curbing excessive exercise behavior. Additionally, the number of self-reported rest days increases from 0.8 to 2.1 days, marking a 162.5% rise. This demonstrates that individuals begin to prioritize rest and recovery, which is crucial for preventing exercise addiction. This work not only provides a new technical means for the early identification of exercise addiction, but also highlights the substantial potential of artificial intelligence in fostering healthy behavior changes.