tambaseddu/ssa-syndrome-surveillance

my first syndrome surveillance system in sub-Saharan Africa # 🌍 Sub-Saharan Africa Syndromic Surveillance System An end-to-end Machine Learning pipeline designed to detect early localized outbreaks of Influenza-Like Illness (ILI) in Sub-Saharan Africa using non-traditional datasets (Community Health Worker mobile reports, transactional proxies, and real-world satellite weather inputs). ## 🚀 How It Works Instead of waiting 1–2 weeks for official clinical diagnoses, this system analyzes early syndromic data. It implements an unsupervised **Isolation Forest** machine learning model to distinguish between normal seasonal illness patterns (such as temperature-driven flu trends or rainfall-driven malaria spikes) and statistical anomalies. ### System Architecture 1. **Weather Fetching:** Pulls real-world weekly climate variables (precipitation and mean temperature) for targeted coordinates using the Open-Meteo API. 2. **Imputation:** Heals missing data windows caused by regional cellular/network dropouts. 3. **Feature Engineering:** Computes rolling historical baselines and climate lag features. 4. **Anomalous Detection:** Trains an unsupervised Isolation Forest model to detect multivariate outliers. 5. **Interactive Dashboard:** Visualizes alerts on a geospatial map and timeline using Streamlit. --- ## 🛠️ Local Setup Instructions ### 1. Clone the Repository & Install Dependencies Ensure you have Python 3.13+ installed. Navigate to your project folder and run: ```bash pip install -r requirements.txt