Federated Adaptive Epidemiological Learning (FAEL): A Novel AI Framework for COVID-19 Pandemic Preparedness in Africa

Background The COVID-19 pandemic exposed critical vulnerabilities in African epidemiological surveillance systems while highlighting the imperative for AI solutions that preserve digital sovereignty. Existing models fail to address the dual challenge of achieving high predictive performance while maintaining data locality—a critical requirement for African countries concerned about technological dependence. Methods We developed Federated Adaptive Epidemiological Learning (FAEL), the first federated learning framework specifically designed for epidemiological surveillance in Africa. Using comprehensive COVID-19 data from all 47 African countries (March 2020–December 2024), we implemented a novel adaptive aggregation algorithm that weights contributions based on data quality, epidemiological similarity, and infrastructure capacity. We validated FAEL through temporal cross-validation across four pandemic waves and geographic cross-validation using leave-one-region-out methodology, comparing against five established baseline models. Results FAEL achieved superior performance across all metrics, with an overall R² score of 0.847 versus 0.792 for the best baseline (LSTM centralized), representing 15.2% improvement (p < 0.001). Geographic validation demonstrated exceptional generalization with R² scores of 0.789–0.845 across African regions. FAEL maintained 84% performance with 40% missing data and achieved epidemic detection 4.2 days earlier than conventional models. Critically, 95.2% of sensitive data remained local, achieving a digital sovereignty index of 0.78 versus 0.45 for centralized approaches. Conclusions FAEL demonstrates that federated learning can exceed centralized model performance while preserving digital sovereignty, offering African countries a viable pathway toward autonomous yet collaborative epidemiological surveillance. The framework's robust performance across diverse African contexts establishes new standards for privacy-preserving AI in global health.