Research map / Vision-language and multimodal / CLIP and zero-shot
SleepLM: Natural-Language Intelligence for Human Sleep
Zongzhe Xu, Zitao Shuai, Eideen Mozaffari, Ravi Shankar Aysola, Rajesh Kumar, Yuzhe Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add signal reconstruction loss (ablation) | SleepLM (ReCoCa) · sleep staging | AUC: 83.9 -> 85.4 | Improved |
| add signal reconstruction loss (ablation) | SleepLM (ReCoCa) · sleep event localization | IoU: 29.6 -> 30.4 | Improved |
| add channel-level captions to local and global captions (ablation) | SleepLM · sleep staging | AUC: 84.6 -> 85.4 | Improved |
| add channel-level captions (ablation) | SleepLM · sleep event IoU | IoU: 30.2 -> 30.4 | Improved |
| add reconstruction to CoCa (ReCoCa vs CoCa) (ablation) | CoCa · text-to-signal retrieval | R@1: 77.5 -> 82.5 | Improved |
| add reconstruction to CoCa (ablation) | CoCa · channel statistics | sMAPE: 7.34 -> 3.15 | Improved |
| add contrastive objective to captioning (CoCa vs Cap) (ablation) | Cap · sleep event IoU | IoU: 29.2 -> 27.4 | Hurt |
| replace captioning-only with ReCoCa (ablation) | Cap · sleep staging | accuracy: 72.0 -> 76.9 | Improved |
| increase model size from tiny to base (38M to 410M) | SleepLM · zero-shot tasks | multiple | Improved |
| multi-source pretraining (SHHS+MrOS+CCSHS) vs SHHS only | SleepLM · zero-shot evaluation | multiple | Improved |
| use SleepLM encoder with linear probe instead of MAE/SimCLR | sleep encoder · few-shot sleep staging WSC | AUROC: approximately 0.90 AUC at 50 samples per class | Improved |
| replace frozen-LLM/fine-tuned VLM fusion with SleepLM | Qwen3-VL-8B-Instruct · sleep staging | AUC: 70.2 -> 85.4 | Improved |
Closest papers by meaning
- SleepFM: Multi-modal Representation Learning for Sleep Across Brain Activity, ECG and Respiratory Signals (ICML 2024 · similarity 0.89)
- Robust Sleep Staging over Incomplete Multimodal Physiological Signals via Contrastive Imagination (NeurIPS 2024 · similarity 0.87)
- Generalizable Sleep Staging via Multi-Level Domain Alignment (AAAI 2024 · similarity 0.86)
- SensorLM: Learning the Language of Wearable Sensors (NeurIPS 2025 · similarity 0.85)
- NeuroLM: A Universal Multi-task Foundation Model for Bridging the Gap between Language and EEG Signals (ICLR 2025 · similarity 0.84)
- BrainLM: A foundation model for brain activity recordings (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
