M-Essa2/Malawi-Public-Health-Systems-LLM-Challenge

AI challenge to build an open‑source LLM assistant for Malawi’s IDSR framework. Fine‑tuned on national technical guidelines, the model supports healthcare workers with accurate, context‑aware disease surveillance, outbreak response, and training, enhancing public health resilience. # 🇲🇼 Malawi Public Health Systems LLM Challenge ### Building an AI Assistant for Malawi’s Integrated Disease Surveillance and Response (IDSR) > Training an open-source Large Language Model (LLM) to support healthcare professionals in Malawi with accurate, context-aware public health guidance. --- ## 🔗 Competition Link Official Zindi Competition Page: zindi.africa --- ## 📌 Overview Malawi follows the **World Health Organization (WHO) Integrated Disease Surveillance and Response (IDSR)** framework to strengthen disease monitoring and response systems. Healthcare professionals across Malawi — including nurses, doctors, researchers, and public health officers — rely on the **Malawi Technical Guidelines (TGs) for IDSR** to: - Identify disease case definitions - Report notifiable conditions - Conduct surveillance activities - Respond to outbreaks and epidemics - Ensure proper data collection and management This challenge, organized by: - **AI Lab at Malawi University of Business and Applied Sciences** - **Public Health Institute of Malawi** aims to develop an AI assistant capable of accurately answering questions about Malawi’s public health system using an open-source LLM trained on the Malawi TGs for IDSR. --- ## 🎯 Challenge Objective The goal is to: - Train an open-source Large Language Model (LLM) - Fine-tune it using a dataset derived from Malawi’s TGs for IDSR - Enable context-specific question answering - Improve accuracy, reliability, and domain understanding - Contribute to the development of the **IntelSurv App** The final solution will serve as: - 📘 An interactive training resource - ⚡ A real-time guidance assistant - 🏥 A decision-support tool for healthcare professionals - 📊 A surveillance enhancement system --- ## 🧠 Technical Approach This project focuses on building a domain-specialized LLM using: - Open-source LLMs (e.g., LLaMA, Mistral, Falcon) - Fine-tuning or paramete …