Training AI to identify diagnostic criteria and treatment methods for advanced and delayed circadian sleep disorders in humans.

ObjectivesThe objective of this review was to evaluate the diagnosis and treatment of advanced and delayed sleep-wake phase disorders (ASWPD, DSWPD), both forms of circadian rhythm sleep-wake disorders (CRSWDs) using both human reviewers and ChatGTP to summarize relevant content. MethodsPubMed-Medline, EbscoHost, and Web of Science were searched for peer reviewed articles. Original research articles published in English were searched for full-text studies. Studies that reported quantitative data on ASWPD and DSWPD diagnosing tools and treatment options in individuals of all ages were assessed. We assessed ChatGTP 4.0s capacity to extract and summarize data from these studies and evaluated it alongside human reviewers. ResultsOur review of 49 articles on CRSWD from the past 20 years found that 91% focused on DSWPD. The most common diagnostic tools were the MEQ, MCQ, Pittsburgh Sleep Quality Index, and Epworth Sleepiness Scale, with primary methods including actigraphy and dim-light-melatonin-onset. Melatonin, often combined with light therapy and CBT, was the predominant treatment. ChatGPT-4.0 facilitated the review process with about 92% accuracy but required manual oversight for optimal results. ConclusionsOur review identified key diagnostic tools, such as MEQ and MCQ surveys, and common treatment options for ASWPD and DSWPD, including melatonin, light therapy, and CBT. The findings underscore the need for comprehensive and individualized treatment approaches for CRSWDs, with a particular focus on expanding research on ASWPD and increasing data representation across age groups. While ChatGPT shows potential in streamlining data extraction for CRSWDs, it still requires human oversight to ensure accuracy.