LilithaMhle/NHI-Project

This repository contains the National Health Insurance (NHI) – Machine Learning Project for the Technical Programming 2 module. The project focuses on predicting healthcare service demand in South Africa under the NHI system using real-world clinical data from the MIMIC-III Clinical Database (Demo Version 1.4). # National Health Insurance (NHI) – Machine Learning Project **Module:** Technical Programming 2 **Due Date:** 30 August 2025 **Group Members (Alphabetical Order):** - Mhle L. – 22322987 - Mncwango A. S. – 22334567 - Msane Z. N. – 22415488 - Mthembu S. H. – 22337669 - Ngwadla M. – 22310899 - Shangase S. – 22315517 --- ## Project Overview This project focuses on predicting healthcare service demand under South Africa's National Health Insurance (NHI) system using real-world clinical data from the **MIMIC-III Clinical Database (Demo Version 1.4)**. The goal is to build a machine learning model where we predict healthcare service demand under the NHI system, using patient demographics, admissions, diagnoses, and treatment data. ## Features - **Machine Learning Life Cycle:** Includes problem definition, data collection, preparation, EDA, feature engineering, model building, and evaluation. - **Data Source:** MIMIC-III Demo dataset, multi-relational with tables including patients, admissions, diagnoses, procedures, prescriptions, lab events, and transfers. - **Predictive Modeling:** Random Forest classifier predicts healthcare service type (Emergency, Elective, Urgent) based on patient demographics and clinical features. - **Data Insights:** Exploratory data analysis visualizes patient demographics, length of stay, admission types, and common diagnoses. - **Feature Engineering:** Includes patient age, number of diagnoses, number of procedures, and admission type. - **Streamlit Dashboard:** Interactive dashboard for exploring data, model results, and application settings. --- ## Streamlit Dashboard – User Interface The **NHI Dashboard** provides an intuitive interface to explore healthcare data and model predictions. It has a **sidebar menu** and four main tabs: - **Home** – Overview of the project, its goals (predicting healthcare demand, forecasting costs, evaluating access disparities), and team members. - **EDA** – Explore datasets with tables, charts for …