AI- Driven Prediction of Lassa Fever Using Evolutional and Random Forest: A Machine Learning Approach for Enhanced Surveillance in West Africa

Lassa fever, a viral hemorrhagic fever endemic to West Africa, poses significant public health challenges with annual case estimates ranging from 100,000 to 300,000 infections and mortality rates reaching 15-20% in hospitalized patients. Current surveillance systems rely predominantly on passive case detection and laboratory confirmation, often resulting in delayed outbreak identification and response. The complex interplay of environmental, climatic, and demographic factors influencing Lassa fever transmission patterns necessitates sophisticated predictive modeling approaches that can process multiple data streams and identify early warning signals for potential outbreaks. This study aims to develop and evaluate an AI-driven prediction model for Lassa fever outbreaks by integrating evolutionary algorithms and Random Forests for optimal feature selection and ensemble learning to enhance early detection and support proactive public health interventions. We implemented a hybrid machine learning approach combining genetic algorithms Random Forest for feature optimization with XGBoost for model training. Evolutionary algorithms and Random Forest were employed to identify the most predictive feature subsets, followed by training and validating an XGBoost model using stratified cross-validation and temporal holdout. The evolutionary algorithm + correlation filter approach achieved exceptional performance with 80.04% accuracy, 61.02% macro precision, and 78.29% weighted F1-score, demonstrating significant improvement over traditional Random Forest feature selection (76.73% accuracy). The model's high accuracy and interpretability make it suitable for integration into existing public health infrastructure, potentially reducing outbreak response time and improving resource allocation for preventive interventions in endemic regions.