How vishwanath-06/pulsepod works
PulsePod is an ECG arrhythmia detection system with a Streamlit inference dashboard and a research workflow for preparing data and training CNN, CNN-LSTM, and spiking-neural-network models. The principal user workflow selects a model and test-beat count, loads a pretrained model and test data, then simulates beat-by-beat inference with confidence, uncertainty, plots, and alerts. The graph also preserves documented dataset preparation and model-training stages. Some research and dashboard internals are unsampled; those links are included only where the README or excerpts support them.
Main entry points: models Pretrained Models · app.py Streamlit Dashboard · create_unified_test_set.py Test Set Merger · 03_data_augmentation_stratification.py MIT-BIH Processing
Components 18
ECG Data
- MIT-BIH Processing
research/03_data_augmentation_stratification.py - Augment and Split
research/03_data_augmentation_stratification.py - Unified Test Set
data/unified_test_set - Test Set Merger
data/create_unified_test_set.py
Research Pipeline
- Model Training
research/04_model_training_experiments.ipynb
Models
- Pretrained Models
models - CNN Classifier
app.py - CNN-LSTM Classifier
app.py - Spiking Classifier
app.py
Inference Dashboard
- Streamlit Dashboard
app.py - Test Data Loader
app.py - Model Loader
app.py - Beat Simulation
app.py - Alerts and Logs
app.py - Live Visualizations
app.py
Other components
- Dashboard User
- MIT-BIH Records
- PTB-XL Records
Connections 18
- MIT-BIH Records → MIT-BIH Processing: processes records
research/03_data_augmentation_stratification.py - MIT-BIH Processing → Augment and Split: segments beats
research/03_data_augmentation_stratification.py - Augment and Split → Model Training: prepares training data
- MIT-BIH Records → Test Set Merger: supplies test data
data/create_unified_test_set.py - PTB-XL Records → Test Set Merger: supplies test data
data/create_unified_test_set.py - Test Set Merger → Unified Test Set: writes combined set
data/create_unified_test_set.py - Model Training → Pretrained Models: trains models
README.md - Dashboard User → Streamlit Dashboard: starts simulation
README.md - Streamlit Dashboard → Model Loader: selects model
app.py - Model Loader → Pretrained Models: loads weights
app.py - Streamlit Dashboard → Test Data Loader: loads test data
app.py - Test Data Loader → Unified Test Set: reads test set
- Streamlit Dashboard → Beat Simulation: runs simulation
app.py - Beat Simulation → Alerts and Logs: checks alerts
app.py - Beat Simulation → Live Visualizations: renders results
app.py - CNN Classifier → Pretrained Models: has saved weights
- CNN-LSTM Classifier → Pretrained Models: has saved weights
- Spiking Classifier → Pretrained Models: has saved weights