Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
tags:
|
| 3 |
+
- ecg
|
| 4 |
+
- multi-label-classification
|
| 5 |
+
- medical
|
| 6 |
+
- cardiology
|
| 7 |
+
library_name: tensorflow
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# ECG Multi-Label Classification Model
|
| 11 |
+
|
| 12 |
+
This model performs multi-label classification on ECG signals to detect:
|
| 13 |
+
- Myocarditis
|
| 14 |
+
- Cardiomyopathy
|
| 15 |
+
- Kawasaki Disease
|
| 16 |
+
- Congenital Heart Disease (CHD)
|
| 17 |
+
- Healthy
|
| 18 |
+
|
| 19 |
+
## Model Architecture
|
| 20 |
+
- 1D CNN with 4 convolutional blocks
|
| 21 |
+
- Input: 12-lead ECG (5000 samples × 12 leads)
|
| 22 |
+
- Output: 5 sigmoid outputs (multi-label)
|
| 23 |
+
|
| 24 |
+
## Training
|
| 25 |
+
- Framework: TensorFlow/Keras
|
| 26 |
+
- Optimizer: Adam
|
| 27 |
+
- Loss: Binary Crossentropy
|
| 28 |
+
- Dataset: Pediatric ECG database
|
| 29 |
+
|
| 30 |
+
## Usage
|
| 31 |
+
|
| 32 |
+
```python
|
| 33 |
+
import tensorflow as tf
|
| 34 |
+
from huggingface_hub import hf_hub_download
|
| 35 |
+
|
| 36 |
+
# Download model
|
| 37 |
+
model_path = hf_hub_download(
|
| 38 |
+
repo_id="Neural-Network-Project/ECG-models",
|
| 39 |
+
filename="checkpoint_final.keras"
|
| 40 |
+
)
|
| 41 |
+
|
| 42 |
+
# Load model
|
| 43 |
+
model = tf.keras.models.load_model(model_path)
|
| 44 |
+
|
| 45 |
+
# Predict (input shape: [batch_size, 5000, 12])
|
| 46 |
+
predictions = model.predict(ecg_data)
|
| 47 |
+
```
|
| 48 |
+
|
| 49 |
+
## Classes
|
| 50 |
+
0. Myocarditis
|
| 51 |
+
1. Cardiomyopathy
|
| 52 |
+
2. Kawasaki Disease
|
| 53 |
+
3. CHD
|
| 54 |
+
4. Healthy
|
| 55 |
+
|
| 56 |
+
## Citation
|
| 57 |
+
Please cite this model if you use it in your research.
|