NIDRA

Sleep stage classification is a cornerstone of sleep research. It relies on resource-intensive manual annotation and is subject to considerable inter-rater variability. While many sophisticated machine learning algorithms exist to solve this task, their adoption in practice has been limited by technical inaccessibility, or restrictive licensing. We present NIDRA: a free, open-source, easy-to-use software platform that makes powerful, validated, state-of-the-art sleep stage classification available to all researchers. NIDRA can be used through a graphical user interface or Python API. NIDRA makes use of existing, powerful machine learning models (e.g., U-Sleep 2.0, and ezscore-f) for polysomnography (PSG) sleep recordings, as well as models fine-tuned for wearable EEG devices such as ZMax, CGX, or DCM. NIDRA can generate hypnograms, hypnodensity plots, classifier probability outputs, sleep statistics, and quality-control visualizations. By combining simplicity, high accuracy, and local execution without data-sharing requirements, NIDRA addresses a critical gap in the sleep research toolkit. It has the potential to accelerate scientific workflows, enhance reproducibility, and lower the barrier to automated scoring.