Architecture: 64→128→256→512 filters with residual connections Training: Focal loss for class imbalance Input: Variable-length 12-lead ECG (5-120 seconds at 500 Hz)

Disease Classes

Fulminant/Viral Myocarditis
Acute Myocarditis
Myocarditis Unspecified
Dilated Cardiomyopathy
Hypertrophic Cardiomyopathy
Cardiomyopathy Unspecified
Noncompaction Ventricular Myocardium
Kawasaki Disease
Ventricular Septal Defect
Atrial Septal Defect
Atrioventricular Septal Defect
Tetralogy of Fallot
Pulmonary Valve Stenosis
Patent Ductus Arteriosus
Pulmonary Artery Stenosis
Pulmonary Valve Regurgitation
Mitral Valve Insufficiency
Congenital Heart Malformation
Healthy

Intended Use

⚠️ Research and educational purposes only - NOT for clinical diagnosis

Training Details

Batch Size: 128
Epochs: 17
Loss: Focal Loss (α=0.25, γ=2.0)
Optimizer: Adam (lr=0.0002)

Citation

bibtex

@misc{ecg-classifier-2025,
  author = {Neural-Network-Project},
  title = {ECG Disease Classifier},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/Neural-Network-Project/ECG-Disease-Classifier}
}