Sepsis is a life-threatening condition that requires early detection to improve patient outcomes. This project develops a deep learning model using LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Unit) networks to predict sepsis onset using ICU time-series patient data.
Deep Learning Models: LSTM and GRU architectures for time-series prediction
Feature Engineering: Uses vitals, lab results, and ICU interventions
Time-Series Processing: Handles irregular time intervals
Performance Evaluation: Uses metrics like AUC-ROC, Accuracy, Precision, Recall, and F1-score
Future Enhancements: Integration with real-time ICU systems, federated learning, and Transformer-based models
The dataset includes 48-hour ICU patient time-series data, consisting of:
Vital signs: Heart rate, blood pressure, respiratory rate, temperature, oxygen saturation
Lab results: White blood cell count (WBC), lactate levels, CRP, creatinine, platelet count, procalcitonin
ICU interventions: Mechanical ventilation, vasopressor use
Sepsis label: Binary classification (0 = No Sepsis, 1 = Sepsis)
Pull requests are welcome! For major changes, please open an issue first to discuss the improvements.
This project is licensed under the MIT License - see the LICENSE file for details.
For questions or collaborations, feel free to reach out:
Vikrant Kawadkar
Email: [email protected]
GitHub: ark5234