title: SenuaLab
emoji: 🧠
colorFrom: indigo
colorTo: blue
sdk: static
app_file: index.html
pinned: false
We develop open and reproducible machine-learning systems for EEG and other biomedical time series. Our work connects signal processing, deep learning, careful evaluation, and human-in-the-loop review so that research artifacts can move cleanly from experiments into practical annotation workflows.
Research focus
- EEG and biomedical time-series analysis
- Epilepsy research and interictal epileptiform discharge (IED) detection
- Self-supervised learning and foundation models for neural signals
- Subject-independent evaluation, calibration, and reproducible deep learning
- Privacy-conscious clinical AI and human-in-the-loop annotation tools
Featured work
| Resource | What it provides |
|---|---|
| EEG IED Detection models | Five supported checkpoints, model metadata, aggregate public-benchmark results, and an EEG Annotation Tool model pack |
| EEG IED Detection source | Leakage-safe training and evaluation pipelines, reproducibility documentation, privacy checks, and inference utilities |
| EEG Annotation Tool integration | Our actively developed EEG review and annotation application, with direct support for installing and using the released models |
From models to annotation
Models released through our Hugging Face organization can be installed directly into EEG Annotation Tool. The integration preserves each model's declared channel order, sampling rate, preprocessing contract, input window, and validation-selected operating threshold while keeping a qualified reviewer in the loop.
Data and privacy
We do not publish private hospital EEG recordings, patient identifiers, physician annotations, or patient-level hospital outputs. Public releases are limited to approved source code, model artifacts, public-dataset references, reproducibility metadata, and aggregate public-benchmark results. Dataset and third-party model licenses remain with their original owners and are documented per release.
Responsible use
Our current EEG releases are research software, not medical devices. Automated candidates must not be used as the sole basis for diagnosis or treatment and must be reviewed by a qualified EEG professional against the original recording.