ECG-based Sleep Apnoea (ECG + XAI, Cross-Dataset)

Code and notebooks for MSc dissertation comparing classical ML models on ECG-derived features with SHAP explainability across Apnea-ECG, UCDDB (St Vincent’s), and SLPDB.

Structure

• notebooks/ — analysis notebooks
• src/ — reusable Python modules
• figures/ — final figures only
• tables/ — exported tables (CSV/DOCX)
• data/ and dataset/ — not included; see Data Access

Data Access (PhysioNet)

• Apnea-ECG: https://physionet.org/content/apnea-ecg/1.0.0/
• UCDDB: https://physionet.org/content/ucddb/1.0.0/
• SLPDB: https://physionet.org/content/slpdb/1.0.0/

Please cite Penzel et al., 2000 (Apnea-ECG) and Goldberger et al., 2000 (PhysioNet).

Reproduce

conda env create -f environment.yml conda activate sleep_apnea_ml jupyter lab