Light-based methods for predicting circadian phase in delayed sleep-wake phase disorder

Area of Science:

• Chronobiology
• Sleep Medicine
• Computational Biology

Background:

• Circadian phase prediction methods are established for healthy individuals but their applicability to clinical populations like DSWPD remains unclear.
• Circadian timing significantly impacts functional outcomes in DSWPD patients.
• Accurate circadian phase assessment is crucial for managing DSWPD.

Purpose of the Study:

• To evaluate the generalizability of existing circadian phase prediction methods to patients with DSWPD.
• To compare the performance of a validated dynamic model and a newly developed statistical model in predicting dim light melatonin onset (DLMO) in DSWPD patients.
• To assess the feasibility of using sleep-wake and light data for circadian phase prediction in a clinical DSWPD cohort.

Main Methods:

• 154 DSWPD patients provided approximately 7 days of sleep-wake and light exposure data.
• A dynamic circadian rhythm model, previously validated in healthy individuals, was applied.
• A statistical model using multiple linear regression was developed, incorporating light exposure, sleep timing, and demographic variables.

Main Results:

• Both the dynamic and statistical models demonstrated comparable performance in predicting DLMO.
• The dynamic model yielded a root mean square error of 68 minutes, achieving ±1-hour accuracy in 58% and ±2-hour accuracy in 95% of participants.
• The statistical model showed a lower root mean square error of 57 minutes, with ±1-hour accuracy in 75% and ±2-hour accuracy in 96% of participants.

Conclusions:

• Circadian phase prediction using light and sleep data is a viable approach for DSWPD.
• These predictive models can enhance the screening, diagnosis, and treatment strategies for DSWPD.
• The findings support the use of objective data for personalized circadian rhythm management in sleep disorders.