0329 Pre-Sleep Physiological Patterns Identify Responders to Closed-Loop Auditory Stimulation for Sleep Enhancement

Abstract Introduction Closed-loop auditory stimulation is able to enhance slow wave activity (SWA) on the electroencephalogram (EEG), however individual responses can vary. We investigated whether pre-sleep physiological features could identify who benefits most from the stimulation and investigated relationships between stimulation responsiveness, sleep enhancement, and next-day cognitive performance. Methods Twenty-eight healthy adults (ages 19-40, mean 27±5 years) experiencing mild sleep restriction underwent three visits, consisting of adaptation followed by randomized stimulation and sham nights. Pre-sleep resting EEG and HRV (8 minutes) were captured to establish baseline brain and cardiac activity before lights-off. During sleep, a phase-locked loop detected slow oscillations in the upstate and triggered 50-ms pink noise bursts for the first five hours. Word-pair recall and psychomotor vigilance task were administered at 1, 4, 7, and 10 hours post-wake. Machine learning models utilizing EEG spectral features and transfer learning from pre-trained sleep architectures were fine-tuned to predict responsiveness. The scores were calculated based on a leave-one-subject-out cross-validation approach. Results Auditory stimulation enhanced SWA in 26 out of 28 subjects. Pre-sleep alpha power positively correlated with SWA enhancement (r=0.436, p< 0.01), as did theta power (r=0.329, p< 0.01) and sleep onset latency (r=0.429, p< 0.01). Delta power showed negative associations (r=-0.299, p=0.01). Individuals with higher HRV, longer sleep onset, and who had less slow oscillation during sleep demonstrated more enhancement. Subject-level temporal embeddings from pre-sleep EEG predicted SWA responsiveness with 80% accuracy (AUC=0.93) and memory enhancement with 90% accuracy (AUC=0.92). Physiological responders showed trends toward improved memory retention at mid-day timepoints (4-7 hours post-wake, p=0.07-0.08) compared to non-responders. Also, our observations show that higher pre-sleep heart rate variability correlated with greater SWA enhancement and predicted better sustained attention at 7-10 hours post-wake. Conclusion Pre-sleep features reliably predict individual responsiveness to auditory stimulation and, to some degree, the associated cognitive benefits. Higher HRV, elevated alpha activity, prolonged sleep onset, and a lower number of slow oscillations indicate a greater enhancement of SWA. This helps to identify individuals who need and would benefit most from intervention. Support (if any) Alzheimer's Association, Michael J. Fox Foundation, CurePSP Sleep Contributions to Neurodegeneration Grant Program (SCN-25-1470707).