Neural-Network-Project/ECG-models

ECG Multi-Label Classification Model

This model performs multi-label classification on ECG signals to detect:

• Myocarditis
• Cardiomyopathy
• Kawasaki Disease
• Congenital Heart Disease (CHD)
• Healthy

Model Architecture

• 1D CNN with 4 convolutional blocks
• Input: 12-lead ECG (5000 samples × 12 leads)
• Output: 5 sigmoid outputs (multi-label)

Training

• Framework: TensorFlow/Keras
• Optimizer: Adam
• Loss: Binary Crossentropy
• Dataset: Pediatric ECG database

Usage

import tensorflow as tf
from huggingface_hub import hf_hub_download

# Download model
model_path = hf_hub_download(
    repo_id="Neural-Network-Project/ECG-models",
    filename="checkpoint_final.keras"
)

# Load model
model = tf.keras.models.load_model(model_path)

# Predict (input shape: [batch_size, 5000, 12])
predictions = model.predict(ecg_data)

Classes

0. Myocarditis
1. Cardiomyopathy
2. Kawasaki Disease
3. CHD
4. Healthy

Citation

Please cite this model if you use it in your research.