metadata
language: python
license: mit
tags:
- audio-classification
- heart-sound
- phonocardiogram
- heart-murmur
datasets:
- circor-heart-sound
metrics:
- accuracy
- f1
- auc
Heart Murmur Detection Model: statistical_random_forest_grid_search
Model Description
- Feature Set: statistical
- Model Architecture: random_forest
- Dataset: PhysioNet CirCor DigiScope v1.0.3
- Task: Heart murmur detection (3-class: Present, Absent, Unknown)
Performance Metrics
| Metric | Value |
|---|---|
| Accuracy | 0.5695 |
| F1 Score | 0.5485 |
| Precision | 0.6114 |
| Recall | 0.5695 |
| AUC-ROC | 0.6768 |
Usage
from src.hub import download_model
from src.models.registry import ModelRegistry
# Download model
model_path = download_model("statistical_random_forest_grid_search")
# Load and use
model = ModelRegistry.get_model("random_forest")
model.load(model_path)
predictions = model.predict(X_test)
Citation
@dataset{circor_2022,
title={The CirCor DigiScope Phonocardiogram Dataset v1.0.3},
author={PhysioNet},
year={2022},
url={https://physionet.org/content/circor-heart-sound/}
}