Humphrey-Galiwango99/heart-disease-ml-app

Heart Disease Risk Prediction (Uganda) This project is my Refactory 2025 Capstone, demonstrating the end-to-end data science pipeline: Data → Cleaning → Modeling → Evaluation → Deployment. We use patient health indicators (age, blood pressure, cholesterol, BMI, lifestyle habits) to predict the risk of heart disease. # heart-disease-ml-app Heart Disease Risk Prediction (Uganda) This project is my Refactory 2025 Capstone, demonstrating the end-to-end data science pipeline: Data → Cleaning → Modeling → Evaluation → Deployment. We use patient health indicators (age, blood pressure, cholesterol, BMI, lifestyle habits) to predict the risk of heart disease. # Problem Statement Heart disease is one of the fastest-growing health challenges in Uganda due to urbanization, poor diet, and inactivity. Unfortunately, resources for screening and prevention are limited. This project builds a machine learning model to help health workers identify high-risk patients early, so that preventive care can be prioritized. # Dataset ~56,000 patient records ## Features: Age, Gender, Height, Weight, Blood Pressure (ap_hi/ap_lo), Cholesterol, Glucose Lifestyle (smoking, alcohol, physical activity) #### Target: Presence of heart disease (0 = no, 1 = yes) # Methodology Data Cleaning & Preprocessing Converted age from days → years Computed BMI from height & weight Removed unrealistic blood pressure values Encoded categorical variables # Modeling Logistic Regression (baseline, interpretable) Random Forest (best balance of accuracy & recall) XGBoost (state-of-the-art boosting method) # Evaluation Metrics Accuracy, Precision, Recall, F1-score, ROC-AUC Focus on Recall (Sensitivity) since missing a sick patient is most dangerous # Results Random Forest performed best overall Key predictors: blood pressure, cholesterol, BMI, age, physical activity High recall ensures fewer sick patients are missed # Deployment The trained model is deployed using Streamlit. ### Live App: Heart Disease Predictor Users can enter patient details (age, BP, cholesterol, BMI, lifestyle) and receive a risk prediction instantly. # How to Run Locally 1. Clone repo: `git clone github.com cd heart-disease-ml-app` 2. Install dependencies: `pip install -r requirements.txt` 3. Run app: `s …