Early Sepsis Detection Using Machine Learning
Worked on a healthcare-focused AI project exploring whether machine learning models can help identify early warning signs of sepsis, a serious and potentially life-threatening medical condition. The project connected computer science, clinical data, and explainable AI to support earlier decision-making.
- Explored clinical features, preprocessing, and model training
- Used Logistic Regression, Random Forest, XGBoost, and ensemble voting
- Evaluated models using confusion matrix, ROC/PR curves, and performance metrics
- Applied SHAP explainability to make model predictions easier to interpret
- Presented the project as an AI-in-healthcare research effort