Chronotype-Aware Sleep Optimization: AI-Driven Physical Activity Strategies

Chronotype is crucial in regulating an individual's circadian rhythm and sleep-wake patterns, yet its impact on the relationship between physical activity and sleep efficiency has been largely overlooked. This study examines how taking chronotype into account can improve our understanding of this relationship and enable the development of more effective sleep improvement strategies. Using actigraphy data from the MESA dataset, we categorized participants by chronotype and applied a convolutional neural network (CNN) and Shapley additive explanations (SHAP) analysis to explore nonlinear interactions. Our results showed that without taking chronotype into account, no significant relationship was observed between physical activity and sleep efficiency. However, when chronotype was taken into account, there were strong correlations for evening type individuals (Pearson = 0.799, Spearman = 0.715) and moderate correlations for morning type individuals (Pearson =0.535, Spearman = 0.443). In addition, SHAP analysis revealed that the optimal timing of physical activity to improve sleep efficiency differed by chronotype, highlighting the need for personalized intervention strategies. This study challenges previous research that has reported inconsistent results by demonstrating that a chronotype-aware approach significantly improves the predictability and optimization of sleep outcomes. By integrating CNN and SHAP for model interpretability, these findings provide a foundation for the development of wearable-based real-time sleep optimization systems and personalized non-pharmacological sleep interventions.Clinical Relevance-Insomnia is a critical health issue linked to higher risks of cardiovascular disease, cognitive decline, and metabolic disorders. While pharmacological treatments are commonly used, long-term reliance raises concerns about side effects and dependency, necessitating safer and more sustainable alternatives. This study integrates wearable actigraphy and AI-driven CNN models to systematically assess the impact of physical activity on sleep efficiency within the framework of circadian rhythms. By identifying chronotype-specific activity patterns, this research provides a foundation for personalized sleep interventions, enhancing the efficacy of non-pharmacological treatments and improving overall health outcomes. From a clinical standpoint, these findings enable healthcare providers to implement chronotype-based physical activity guidelines, optimizing behavioral therapies while reducing dependence on medication. Additionally, the use of wearable technology combined with AI offers a scalable, real-time monitoring system that supports individualized treatment approaches. Incorporating these insights into clinical practice and public health initiatives may refine sleep management strategies, promote precision medicine, and advance non-pharmacological treatment paradigms for sleep disorders.