CosmosKyeremeh/malaria-risk-prediction-ghana

# Malaria Severity Prediction Using Supervised Machine Learning Models (Ghana Context) **Status:** 🚧 In development — Phase 3 (Environment & Repository Setup) ## Overview This project explores whether supervised machine learning models can classify malaria case severity — non-malarial infection (nMI), uncomplicated malaria (UM), or severe malaria (SM) — using haematological (blood) parameters, based on data collected in Ghana. It is a student research/portfolio project developed as part of independent study in AI/ML/Data Science, and is informed by a literature review on AI-driven clinical decision support systems in sub-Saharan African healthcare settings (see `docs/`). **This is a research prototype, not a diagnostic or clinical decision-making tool.** It has not been clinically validated and is not intended for use on real patients. See Limitations & Ethical Considerations below. ## Motivation Malaria remains Ghana's leading cause of morbidity, accounting for a large share of outpatient visits and hospital admissions nationally. Most published disease-prediction ML projects rely on datasets collected in Western populations, which raises real questions about whether their findings transfer to Ghanaian clinical contexts. This project uses a dataset actually collected in Ghana, and is built with the long-term (post-graduation, non-clinical) goal of exploring how such tools might one day support — not replace — health workers in resource-constrained clinics. ## Dataset - **Source:** Haematological data from 2,207 participants collected in Ghana (Morang et al.), publicly hosted on Kaggle. - **Classes:** non-malarial infection (nMI), uncomplicated malaria (UM), severe malaria (SM). - **Note on provenance:** Before redistributing or committing any portion of this dataset, its Kaggle license/usage terms must be checked and respected. Raw data is **not** committed to this repository — see `data/raw/README.md` for the download procedure instead. ## Project Structu …