Personalized Closed-Loop Auditory Stimulation of Sleep Spindles: Algorithmic Design and Validation in Young Adults

Abstract

Sleep spindles have been implicated in systems-level memory consolidation, yet closed-loop auditory stimulation protocols remain difficult to personalize to an individual's spindle dynamics. I present a real-time, subject-specific pipeline that detects and phase-locks brief pink-noise bursts to the up-state of fast spindles during N2 sleep. The approach adapts detection thresholds to an individual's sigma peak and morphology, employs drift-robust phase tracking, and gates stimulation by an uncertainty controller to minimize arousals. In a preregistered validation with N=40 healthy adults (18–30 y), the system increased central fast spindle density by 15% relative to sham without altering macro-architecture. Overnight improvement on a probabilistic sequence task correlated with stimulation-induced changes in spindle density (β=0.28, p=.009) after adjusting for baseline performance and total sleep time. These results support individualized, closed-loop targeting of spindles as a feasible route to enhance consolidation and provide a reproducible framework for educational and clinical translation.