Beyond Algorithms: The Case for Standardized Reporting in AI Sleep Scoring.

Recent years have witnessed important developments in AI reporting guidelines across medicine.The TRIPOD-AI statement (2024) provides updated guidance for clinical prediction models using machine learning, with 27 main items (comprising 52 checklist subitems) addressing model transparency, reproducibility, and validation rigor. 7The CONSORT-AI extension (2020) established 14 new reporting items for randomized controlled trials evaluating AI interventions, emphasizing description of AI integration, input data handling, and human-AI interaction. 8The companion SPIRIT-AI guideline addresses trial protocol reporting. 9The STARD-AI extension for diagnostic accuracy studies is currently in development, 10 and the Checklist for AI in Medical Imaging (CLAIM) was initially published in 2020 and updated in 2024, providing specialty-specific guidance for radiology applications. 11,12p Analysis These frameworks represent substantial progress but remain insufficient for sleep medicine applications.Spitschan et al assessed journal-level transparency policies in 28 sleep and chronobiology journals using the TOP Factor, finding median scores of only 2.5 out of 29 points, suggesting limited adoption of reporting standards. 13More recently, Collins