0004 Using Apple Watch to predict circadian phase in night shift workers

Abstract Introduction A critical obstacle for circadian medicine is the lack of feasibility in measuring circadian phase in the clinic. Existing tools such as assessment of dim light melatonin onset (DLMO) are too resource intensive, especially in populations with extreme circadian disruption such as night shift workers. Recent studies have demonstrated the validity and feasibility of estimating circadian phase via mathematical modeling data collected with wearable technology. However, these studies have mostly relied on research grade devices (e.g., actigraphs) that have limited scalability. In this work, we validate the use of a consumer wearable (the Apple Watch) to predict DLMO in a population of night shift workers. Methods A sample of 21 fixed-night shift workers wore an Apple Watch for two weeks before completing DLMO in the lab. DLMO was assessed via hourly salivary melatonin samples collected in dim light (< 10 lux) for a period of 24 hours. Activity data was used as input into the Hannay model of the circadian clock to produce a predicted DLMO, which was then compared to in-lab DLMO. Results Model predictions of DLMO showed high correlation with in-lab DLMO, with a Lin’s concordance correlation coefficient (CCC) of 0.81, with a mean absolute error (MAE) of 2.10. This was comparable to the previously published validation using research grade actigraphy (CCC = 0.70; MAE = 2.88). Conclusion This study is the first to provide evidence suggesting that estimates of circadian phase using the Apple Watch have comparable validity with research grade actigraphy. These results have significant implications for the scalability of circadian medicine, particularly as consumer-based wearable technology is already commonplace. Apple Watches are also the most common consumer-based wearable device. Future research should extend findings to other wearable devices, especially devices at a lower price-point to increase accessibility. Support (if any) Support for this study was provided from the American Academy of Sleep Medicine Foundation (245-SR-21) awarded to Dr. Philip Cheng.