Traditional exercise and health management systems often rely on static rules and generalized protocols. These approaches make it difficult to integrate multi-source heterogeneous data or adapt dynamically to individual needs. As a result, user engagement remains low and intervention effectiveness is limited. To address these challenges, this study developed a Generative Artificial Intelligence (GAI)-driven personalized management system. The system dynamically generates exercise prescriptions, nutritional recommendations, and risk alerts. It integrates several advanced components: a Transformer-based Large Language Model (LLM) for semantic profiling, a Generative Adversarial Network (GAN) for synthetic data augmentation, and a Graph Neural Network (GNN) to capture complex relationships among health factors. A Deep Reinforcement Learning (DRL) framework is then applied to optimize adaptive decision-making. Empirical results showed strong performance. The system achieved 92.7% accuracy in exercise type recommendations, a calorie prediction error of only 7.9%, and a personalized recommendation rating of 4.7 out of 5. User satisfaction increased by 35.3%, and significant improvements were observed in key health indicators, including BMI and resting heart rate. Overall, this study established an efficient and scalable technological foundation for intelligent health management. The findings highlight the potential of GAI to enable precise and effective health interventions.
Wang et al. (Thu,) studied this question.