0544 Novel Continuous Respiratory Metrics Predict Sleep Instability Across a Large Clinical Sleep Cohort

Abstract Introduction Traditional sleep apnea metrics (e.g. AHI) quantify discrete respiratory events but fail to capture qualitative components of breathing physiology. Our group has developed two novel respiratory pathophysiological biomarkers: Self Similarity (SS) and the Breathing Stability Index (BSI), which quantify respiratory waxing/waning and breath-to-breath stability across the night, respectively. We hypothesized that these measures of respiratory instability will correlate with and predict broader measures of sleep instability. Methods We analyzed 24,596 polysomnograms from two large academic medical centers, including diagnostic (n=13,579), split-night (n=5,185), and titration (n=5,832) studies. We compared participants stratified to the highest versus lowest quintile of each respiratory instability marker across sleep variables using two-sample t-tests and Chi-square tests, with effect sizes for interpretable magnitude. A correlation matrix was generated with key sleep-quality variables (N1% sleep, sleep efficiency, wake-after-sleep-onset (WASO), and sleep fragmentation index (SFI)). Finally, we assessed predictive capacity by fitting linear regression models between respiratory instability markers and sleep-quality outcomes. Results High SS-indices were associated with poorer sleep quality with higher N1% (d=0.58), lower N2%/N3% (d=-0.13/-0.39), more WASO (d=0.38), lower sleep efficiency (d =-0.36), and higher SFI (d=0.54) than low SS indices. High BSI’s showed similar association across sleep quality markers. The SS-Index was highly associated with BSI (r=0.43), as well as N1% (r=0.24) and SFI (r=0.16). BSI was similarly correlated with N1% (r=0.24) as well as SFI (r=0.23), sleep efficiency (r=-0.15), and WASO (r=0.15). A combined linear regression model (SS+RSI) best predicted sleep quality outcomes: sleep efficiency (adjR2=0.095), N1% (adjR2=0.113), and SFI (adjR2=0.111). Both SS and BSI contributed unique, non-redundant information to the models. Age-interaction terms indicated that adverse impacts of respiratory instability on sleep quality increases with advancing age. Conclusion Application of two novel respiratory stability markers (SS and BSI) in a large clinical dataset confirmed that high respiratory instability is consistently associated with a characteristic pattern of light, fragmented sleep. SS and BSI correlate moderately with each other but contribute distinct, complementary information. Both SS and BSI independently captured variance related to sleep fragmentation, N1%, WASO, and sleep efficiency and may provide additional information beyond the AHI to guide targeted sleep apnea therapies. Support (if any) 1R01HL161253-01A1