Artificial intelligence and sleep medicine II: A scoping review of applications, advancements, and future directions.

BACKGROUND

Artificial intelligence (AI) has rapidly advanced in sleep medicine, offering innovative solutions for sleep disorder management. The integration of AI technologies presents unprecedented opportunities to address longstanding challenges in diagnosis, monitoring, and treatment.

METHODS

This scoping review summarizes current AI applications in sleep medicine following PRISMA-ScR. A comprehensive search was conducted across various databases from inception to 2024, following PRISMA-ScR guidelines. Eligible studies included original AI research, systematic reviews, and meta-analyses focusing on sleep medicine.

RESULTS

AI models for obstructive sleep apnea (OSA) detection demonstrated accuracy between 85 % and 99 % in controlled settings, with outlier up to 99.9 %. Deep learning approaches for sleep staging and quality assessment match human expert performance. AI applications have extended to other disorders like insomnia and narcolepsy. Wearable technologies and non-contact sensors enable continuous monitoring outside laboratory settings, with some achieving sensitivity >96 % and specificity >97 % for OSA detection. Significant real-world validation gaps persist.

CONCLUSIONS

While AI shows transformative potential in sleep medicine, implementation challenges remain. Future research should prioritize multicenter validation, ethical governance, and culturally inclusive AI frameworks to ensure safe and equitable clinical translation. Addressing ethical considerations and standardizing methodological reporting are essential for widespread adoption of these technologies in clinical practice.