Optimizing auditory stimulation timing in NREM sleep using brain–heart rhythms: continuous phase analysis and multidimensional phase-locking

Study Objectives: Auditory stimulation during non-rapid eye movement sleep effectively enhances slow oscillations and slow-wave activity (SWA) when precisely timed to certain phases of the slow oscillation. However, timing precision remains a core challenge. Recent evidence suggests that heart rate components may provide effective complementary timing cues. This study examined which heart rate phases are associated with stronger stimulation responses using continuous phase analysis and evaluated a multidimensional phase-comparison approach that integrates heart rate and EEG slow oscillation phases. Methods: Polysomnography recordings from 133 adolescents were analyzed. Auditory tones were delivered randomly every 15–30 s during non-rapid eye movement (NREM) sleep. Instantaneous phases of EEG slow oscillation (~0.8 Hz) and heart rate components in the low-frequency (0.04–0.15 Hz) and high-frequency (0.15–0.4 Hz) bands were extracted for continuous phase analysis. Tone-evoked slow oscillation amplitude and slow-wave activity were further compared across three phase-locking strategies: unimodal (slow oscillation-only or heart rate-component-only) and combined (EEG–heart rate). Results: Responses were largest when tones occurred near the heart rate-low-frequency up-peak and heart rate-heart rate down-peak. Phase analyses showed that tones occurring at the optimal heart rate component phases were accompanied by increases in slow oscillation amplitude by up to ~22 μV and SWA by 12 per cent, indicating that peripheral signals can serve as strong, independent timing cues. Slow oscillation-only phase-locking also produced notable effects (~18 μV slow oscillation amplitude, 19% SWA increase). Combining slow oscillation and heart rate phases yielded the greatest effects, with increases of ~38 μV in slow oscillation amplitude and 32 per cent in SWA. Conclusions: Oscillatory phases derived from heart rhythms provide effective timing information that may be useful for closed-loop auditory stimulation and reflect brain–heart coupling during sleep. A multidimensional phase-based approach that integrates EEG slow oscillations with instantaneous heart rate phases may support more precise control and stronger enhancement of deep sleep than unimodal approaches, suggesting a new framework for closed- loop neuromodulation.