Sleep assessment technologies for premature infants: a comprehensive review of traditional methods and emerging innovations

Abstract This study comprehensively reviewed sleep assessment methodologies for premature infants, focusing on both traditional methods and emerging innovations. It systematically analyzed the limitations of conventional approaches, such as behavioral observation and polysomnography, and explored the potential of artificial intelligence and multimodal physiological signal fusion to improve sleep monitoring accuracy. The integration of heart rate variability, respiratory pattern, and brain activity through machine learning algorithms demonstrated promising results in distinguishing sleep-wake states. However, challenges persisted, including limited sample sizes and the need for large-scale clinical validation. The study highlighted the importance of developing noninvasive, real-time monitoring systems to reduce interference and improve sleep quality in preterm infants, ultimately promoting their neurodevelopmental outcomes. Future research directions include optimizing artificial intelligence–driven technologies, enhancing noninvasive monitoring tools, and constructing user-friendly visualization platforms for clinical applications.