1274 Practical Considerations for Using Pulse Oximetry Data from Smartwatches to Promote Self-Management of OSA

Abstract Introduction Consumer-grade smartwatches with pulse oximetry are increasingly available. We previously demonstrated in a pilot randomized trial (NCT06039865) that weekly, patient-facing reports summarizing nightly minute-to-minute pulse oximetry (SpO2) data collected from smartwatches improved PAP usage in patients with OSA who were non-adherent to PAP therapy. In this study, we further analyzed the trial’s pulse oximetry data to describe completeness of SpO2 data and patterns of desaturations to inform the feasibility of the intervention and future applications of this novel use of smartwatches. Methods Participants were instructed to wear a smartwatch (Fitbit Charge 5) for 28 consecutive days. Smartwatch data were sent to a database that the study team downloaded weekly. The number of days (out of 28) with available SpO2 data, total minutes SpO2 collected per sleep period, and minutes of SpO2 < 88% were analyzed. Possible explanations for data loss were enumerated based on the research team’s troubleshooting efforts during the intervention. Results Data from 49 participants (mean age 47.8 years; 12% female) were available. The number of days with available SpO2 data per participant ranged from 11 to 28 days out of the 28-day study period. The average daily total minutes of SpO2 collected per sleep period across participants ranged from 1.6 to 6.9 hours (96 to 416 minutes), with night-to-night variability within each participant as well. Explanations for incomplete data included: forgetting to wear the smartwatch, obstructions to the smartwatch SpO2 sensing (e.g., tattoos), and participants not meeting the smartwatch’s internal requirements for sleep to begin the minute-by-minute SpO2 sampling. 26 (53%) participants had at least 1 sleep period with SpO2 < 88% for ≥ 10 minutes; % sleep periods with SpO2 < 88% for ≥ 10 minutes mean was 8.6% (SD 6.5; min 3.7%, max 30.4%). Conclusion Minute-to-minute SpO2 data collected by smartwatches among patients nonadherent to PAP therapy show a wide range of values. Determining whether there is a minimum amount and appropriately distributed coverage of SpO2 data needed for patient facing SpO2 reports to significantly alter PAP adherence would provide clinicians a way to leverage data provided by smartwatches. Support (if any) ResMed Foundation, Google