Momahmoses/malaria-risk-prediction

ML model predicting malaria risk in Nigeria. # Malaria Risk Prediction for Nigeria ML classification model predicting malaria risk at the individual and community level across Nigeria, supporting NMEP and state health ministries in targeting prevention resources to highest-risk populations. --- ## Problem Statement Nigeria accounts for 27% of global malaria deaths. Limited diagnostic capacity means many cases go undetected until severe. This model flags high-risk individuals based on environmental, demographic, and behavioural features, enabling pre-emptive bed net distribution, IRS targeting, and IPT prioritisation. --- ## Features | Feature | Description | |---------|-------------| | Multi-Model Training | Random Forest, Gradient Boosting, Logistic Regression, Decision Tree | | Cross-Validation | Stratified K-fold with AUC, F1, and accuracy reporting | | Feature Importance | Environmental and demographic risk driver ranking | | AUC-ROC Analysis | Threshold optimisation for field deployment | | Model Serialisation | Best model exported for integration with DHIS2 | --- ## Tech Stack | Layer | Technology | |-------|-----------| | Machine Learning | scikit-learn | | Data | pandas, NumPy | | Visualisation | Matplotlib, Seaborn | | Serialisation | joblib | --- ## Quick Start ```bash git clone github.com cd malaria-risk-prediction pip install pandas scikit-learn matplotlib seaborn joblib numpy python data_generator.py python train.py python predict.py ``` --- ## Data Sources - NMEP malaria indicator survey data - NIMET rainfall and temperature records - WorldPop gridded population data - DHS Programme Nigeria survey --- ## Author **Momah Moses**, Geospatial AI Engineer & Data Scientist GitHub · Portfolio