Human-Centered AI in Sleep Health Management: Scoping Review of Stakeholder Perspectives and Co-Design Practices

Area of Science:

• Sleep medicine
• Artificial Intelligence (AI)
• Human-Centered AI (HCAI)

Background:

• Sleep disorders pose significant public health challenges, linked to cardiovascular and neurocognitive issues.
• AI offers personalized sleep medicine potential, but clinical integration is hindered by a lack of human-centered design and stakeholder engagement.
• Current research often prioritizes AI algorithms over usability and patient trust.

Purpose of the Study:

• To systematically map human-centered AI (HCAI) research in sleep medicine.
• To evaluate stakeholder involvement (patients, clinicians, technologists) in AI tool design, validation, and implementation.
• To identify research gaps and trends in HCAI for sleep health.

Main Methods:

• Scoping review following PRISMA-ScR guidelines.
• Searched 8 databases for literature up to June 18, 2026.
• Included 34 studies on AI for sleep health with human-centered components, categorized by AI type and stakeholder engagement.

Main Results:

• Generative AI (GenAI) research focuses on output accuracy, lacking patient participatory design.
• Deep learning research emphasizes explainable AI for clinicians but lacks clinical implementation.
• Mobile health and wearables show a balanced HCAI ecosystem with full translational cycle.
• AI is shifting towards interactive therapeutic agents, with potential for empathetic user perception.

Conclusions:

• Findings highlight methodological disparities and the need for upstream participatory design, especially for GenAI.
• Standardized protocols, transparency, and bias mitigation are crucial for clinical AI adoption in sleep medicine.
• This review systematically maps sociotechnical factors, unlike others focusing solely on algorithmic performance.