0315 Exploring AHI Variability Across Nights: Insights from Wearable Technology in Patients with No or Mild OSA

Abstract Introduction Night-to-night variability (NtNV) of obstructive sleep apnea (OSA) severity is a widely recognized phenomenon in sleep testing. Studies suggested that 20% of patients with mild-to-moderate OSA may be misdiagnosed or misclassified on the first night. Multiple-night assessments utilizing wearables may present a convenient approach to enhance the reliability of OSA estimation. The Belun Sleep System (BLS-100, a.k.a., Belun Ring) is an FDA-cleared, deep-learning-powered home sleep apnea testing system. This analysis aims to assess the NtNV of Belun apnea-hypopnea index (bAHI, events/h) in subjects with no or mild OSA. Methods Participants were recruited from a dental school orthodontic clinic for multiple-night Belun Ring (BR) testing. The bAHI was derived for each night, and a reference bAHI was computed based on the mean from all nights. Linear mixed-effects (LME) models were employed to evaluate differences between night-specific bAHI (1st, 2nd, 3rd night, etc) and the reference bAHI. Results Thirty-six patients underwent BR testing, with 30 completing ≥3 nights with ≥60 mins/night. Analysis was conducted on 28 subjects with no or mild OSA: age 31.7; 73% female; BMI 24.6; average 6.5 nights (3-10); mean total sleep time 340.0 mins; average AHI 6.2; 46% no OSA and 54% mild OSA. No significant differences were observed between the first night bAHI and the reference bAHI (6.8 vs. 6.2, P=0.362). This trend held true for subgroups with no OSA (3.9 vs. 3.6, P=0.712) and mild OSA (9.3 vs. 8.5, P=0.716). The first night study misclassified 4 (14%) subjects, including 3 (10%) non-OSAs as mild OSA and 1 (4%) mild as moderate OSA (κ=0.72, P 0.001). While the number of assessment nights was non-significant in NtNV (LME, P=0.618), a drop in misclassification from 14% to 4% after 4-night (κ=0.92, P 0.001) suggested that a 4-night evaluation may offer a more dependable bAHI assessment. Conclusion This research marks the first exploration into the AHI NtNV utilizing wearables in subjects with no or mild OSA. No significant bAHI differences were observed across the nights. A 4-night assessment may provide a more reliable OSA estimation within this particular population. Further investigation into AHI NtNV utilizing BR in moderate-to-severe OSA is warranted. Support (if any)