🫀 CardioVision AI — ECG Arrhythmia Classifier

Upload a single heartbeat (187 time samples in CSV format).
Model trained on MIT‑BIH Arrhythmia Database — 91% Accuracy.

Examples

📖 How to Use

  1. Prepare a CSV file with exactly 187 comma-separated values representing one heartbeat.
  2. Upload the file or use the sample below.
  3. View prediction probabilities and signal plot.

🔬 Model Details

  • Architecture: ResNet18‑1D + LoRA (0.4% trainable parameters)
  • Loss: Focal Loss with class weights
  • Explainability: Grad‑CAM (available in GitHub repo)

⭐ GitHub Repository · 📚 MIT‑BIH Dataset