Reliability and validity of a context-aware sleep holistic cycle gate algorithm for personalized digital healthcare and sleep prediction

This study evaluated the reliability and validity of a context-aware sleep holistic cycle gate algorithm and explored its potential for artificial intelligence (AI)-based personalized sleep management in digital health-care. A 6-week prospective observational study was conducted in 25 adults aged 20–65 years. Participants recorded daytime activities, eve-ning behaviors, presleep environmental factors, sleep outcomes, and feedback adherence using a mobile application. Five sleep-related indices were calculated, and a context-aware gate algorithm comprising five domains was applied. Reliability was assessed through internal consistency and reproducibility analyses, while validity was examined using correlations with sleep quality indicators and receiver operating characteristic curve analyses. Data completion and adherence rates were 80.3%±9.8% and 72.7%±11.6%, respectively. The sleep readiness index was positively correlated with subjective sleep quality (r=0.53, P<0.01). The algorithm demonstrated moderate predictive performance (area under the curve=0.73), with the environment and presleep gates showing the highest accuracy. Following personalized feedback, sleep onset latency decreased by 12.6%, and perceived sleep quality improved significantly (P<0.05). In conclusions, the context-aware sleep holistic cycle gate algorithm demonstrated acceptable reliability and validity for predicting sleep quality. These findings suggest that the proposed framework may serve as a foundational platform for the development of AI-driven personalized sleep management systems and future digital healthcare applications related to sleep rehabilitation and health promotion.