Development of an automated REM sleep deprivation device for mice in neuroscience research.

Rapid eye movement (REM) sleep deprivation paradigms are widely used to investigate the functional role of REM sleep, yet conventional methods often disrupt natural housing conditions and introduce stress-related confounds. Here, we present an open-source, home-cage-compatible system for automated REM sleep deprivation in mice that integrates physiological monitoring with real-time closed-loop motor control. Electroencephalography (EEG) and electromyography (EMG) signals are acquired using an Open Ephys acquisition board coupled to an Intan RHD-series headstage (Intan Technologies, USA). Real-time REM detection relies solely on EEG-derived spectral features, whereas offline brain-state annotation is performed using combined EEG and EMG recordings. The real-time classifier discriminates REM, wake, and non-rapid eye movement (NREM) states with accuracies of 88.6%, 80.4%, and 89.1%, respectively. Upon REM detection, the Open Ephys board outputs a Transistor-Transistor Logic (TTL) signal to a servo motor controller (Maxon ESCON 50/5), which drives a gently moving platform to selectively disrupt REM episodes. During a continuous 48 h deprivation protocol, REM sleep was robustly suppressed. At 24 h, the baseline-to-deprivation ratio for REM was approximately 15.9x, while wake and NREM ratios were approximately 0.66x and 1.46x, respectively. Sustained REM suppression remained evident at 48 h (∼7.9x), demonstrating stable closed-loop performance. All hardware design files and source code are publicly available.