Merryheart/nesa-lassa-fever-surveillance

Patient-level Lassa fever surveillance analysis for Nigeria, 2020–2024. Covers data cleaning, epidemiological analysis, predictive modelling, and Power BI dashboard. Built on a synthetic dataset grounded in real NCDC statistics. # nesa-lassa-fever-surveillance Patient-level Lassa fever surveillance analysis for Nigeria, 2020–2024. Covers data cleaning, epidemiological analysis, predictive modelling, and Power BI dashboard. Built on a synthetic dataset grounded in real NCDC statistics. # NESA Lassa Fever Surveillance Analysis **Nigeria Epidemiological Surveillance Agency (NESA)** — Fictional Organisation > *Protecting Nigeria. Predicting Tomorrow.* --- ## Background Lassa fever remains one of Nigeria's most persistent public health threats, endemic across multiple states with case fatality rates far exceeding WHO's recommended threshold. Every year, Nigeria loses hundreds of lives — many of them preventable. This project asks a specific question on behalf of decision-makers at NESA: **Are patients dying because they arrive too late for treatment to work — or because treatment fails even when they arrive in time?** The answer determines where limited intervention budget goes: community surveillance and early detection, or treatment capacity at facility level. --- ## Project Structure ``` nesa-lassa-fever-surveillance/ │ ├── data/ │ ├── raw/ # Raw dataset as generated │ └── cleaned/ # Cleaned dataset after processing │ ├── notebooks/ │ ├── 00_data_generation.py # Synthetic data generation script │ ├── 01_data_cleaning.ipynb # Data cleaning pipeline │ ├── 02_analysis.ipynb # Epidemiological analysis (5 angles) │ └── 03_ml_model.ipynb # Predictive modelling │ ├── outputs/ # Generated charts and exports ├── dashboard/ # Power BI .pbix file └── README.md ``` --- ## Dataset - **File:** `nesa_lassa_raw_2020_2024.csv` - **Rows:** 100,400 patient records - **Columns:** 51 variables - **Period:** 2020 – 2024 - **Level:** Patient-level surveillance records The dataset is synthetic but epidemiologically grounded — all statistical relationships (CFR by state, seaso …