260 Beyond Validation Testing: A Sleep Tracker for Longitudinal Data Collection in Operational Environments

Introduction Sleep tracking wearables are increasingly being validation tested against polysomnography (PSG) and actigraphy, but they may not be ideal for long-term epidemiological sleep studies, or for use in operational environments. A given device’s short battery life, limited data storage capacity, inability to detect naps or estimate sleep architecture, or privacy concerns may discourage researchers from using wearables to collect objective sleep data in real-world settings. The Zulu watch (Institutes for Behavior Resources) is designed to collect longitudinal sleep data in operational populations with irregular sleep patterns, such as long-haul pilots or shift workers. It is capable of on-wrist sleep-wake determination, nap detection, on-wrist sleep depth scoring (i.e., interrupted sleep, light sleep, or deep sleep), on-wrist data storage up to 80 sleep intervals, and year-long battery life. Laboratory testing is an important initial step toward establishing the performance of a device for longitudinal real-world sleep evaluation; therefore, the Zulu watch sleep tracking was subjected to testing against gold-standard PSG and actigraphy. Methods Eight healthy young adult participants (30.4±3.2 years; mean±SD) wore a Zulu watch and Philips Respironics Actiwatch 2 simultaneously over a 3-day laboratory PSG sleep study, with 8 hours time-in-bed each night. Overall epoch-by-epoch agreement of sensitivity (for sleep), specificity (for wake), and accuracy of Zulu watch data were tested against PSG and Actiwatch 2. Results Compared with either PSG or actigraphy, both accuracy and sensitivity for Zulu watch sleep-wake determination were >90% while specificity was low (~26% vs. PSG, ~33% vs. actigraphy). Accuracy for sleep scoring vs. PSG was ~87% for interrupted sleep, ~52% for light sleep, and ~49% for deep sleep. Conclusion The Zulu watch showed mixed results but may be a viable candidate for sleep evaluation based on initial laboratory performance testing in healthy adults. The next steps will be to compare the Zulu watch against self-report of sleep in operational and substance use disorder populations. Longitudinal epidemiological sleep studies can become more feasible if technology is tailored to the specific needs of the real-world environment. Support (if any) Medical Technology Enterprise Consortium award MTEC-17-08-Multi-Topic-0104; Office of Naval Research, Code 34.