0AC-II/ndhs-hypertension

Machine learning model predicting hypertension risk among Nigerian adults using Nigeria DHS 2024 data. Includes Streamlit web app for public use. # Sociodemographic Predictors of Hypertension Risk Among Nigerian Adults ## A Machine Learning Analysis of Nigeria DHS 2024 > Try the live app: hypertension-risk-calc.streamlit.app --- ## Project Overview This project builds a machine learning model to estimate hypertension risk among Nigerian adults using purely sociodemographic features -- no clinical measurements required. It uses the Nigeria Demographic and Health Survey (NDHS) 2024, the most recent nationally representative health survey available for Nigeria. The core question: **Can we identify which unscreened Nigerians are most likely to be hypertensive, using only information we already know about them?** --- ## Why This Matters Hypertension is the leading modifiable risk factor for cardiovascular disease in sub-Saharan Africa. In Nigeria, a large proportion of hypertensive adults have never been screened. Without a diagnosis, there is no treatment. Without treatment, the condition progresses silently. This model is not a clinical diagnostic tool. It is a **population-level screening signal** -- designed to help health programs prioritize outreach to communities and demographic groups where undetected hypertension is most likely concentrated. --- ## Data Source - **Survey:** Nigeria Demographic and Health Surveys (NDHS) 2024 - **Phase:** DHS-VIII (Standard DHS) - **Files used:** Individual Women's Recode (IR) + Men's Recode (MR) - **Access:** DHS Program (registration required) -- dhsprogram.com - **Combined sample:** 51,254 adults (women 15-49, men 15-59) - **Screened population (model training):** 24,333 adults - **Unscreened population (risk inference):** 26,673 adults --- ## Target Variable - **chd02:** Ever been told you have high blood pressure or hypertension - Applied only to adults who confirmed they had been screened (chd01 = 1) - Positive rate in screened population: **13.7%** --- ## Features Used | Variable | Descriptio …