Brugada Triage CNN: Reproducible 100 Hz ECG Pipeline (PhysioNet Brugada-HUCA)

This record archives the full, reproducible analysis pipeline used for our manuscript on confidence-gated (selective referral) triage of Brugada syndrome from low-resolution resting ECG sampled at 100 Hz using the PhysioNet Brugada-HUCA dataset. What is included:- End-to-end scripts to download the public dataset (without redistributing patient-level ECG data)- Preprocessing to extract V1–V3 median beats at 100 Hz- Compact 1D CNN training and evaluation using stratified 3-fold cross-validation with pooled out-of-fold predictions- Locked triage threshold (tau) and exported metrics- Figure generation (ROC, PR, confusion matrix, calibration) Data availability:This Zenodo record and the linked GitHub repository contain code only. No patient-level ECG recordings are included. Users must obtain the PhysioNet Brugada-HUCA dataset directly from PhysioNet and comply with its terms of use. How to run:See README.md in the repository for step-by-step commands (download → preprocess → train/eval → figures). Version:This archive corresponds to GitHub release v1.0.0.