SLEEPJEV v0.1.0

SLEEPJEV v0.1.0

Initial public research release of SLEEPJEV, an open-source PyTorch framework for reusable long-horizon polysomnography (PSG) representations and runtime semantic sleep analysis.

Highlights

• Reusable overnight PSG encoding
• Runtime sleep staging and sleep-event queries
• JEV-inspired typed decisions: Choice, Noul, and Score
• Sparse temporal retrieval and option-conditioned scoring
• Sleep-EDF / NSRR-style data adapters
• Reproducible high-query workload benchmarks
• Interactive public demo with real SHHS feature-cache replay snapshots

Important scope note

SLEEPJEV is a research prototype, not a medical device or clinical decision system. The public demo is a static replay because GitHub Pages does not run the PyTorch backend. Run python demo/server.py locally for checkpoint-backed inference.