Systems Architecture for Real Time Epidemiological Prediction and Control

Abstract

Emerging contagious diseases create enormous challenges to highly interconnected human societies. Over the next decades, their devastating potential will grow because of pervasive urbanization and globalization, peak populations, aging, climate change, and stresses from uneven development. Progress in computational data sciences provides new opportunities for faster and more effective public health responses but demands new approaches that are inherently statistical and able to learn from diverse streaming data. Here we show how to combine epidemic models with probability theory and Bayesian learning to create statistical frameworks for data assimilation, adaptive prediction and control. Resulting models estimate epidemic parameters in real time, quantifying the comparative success of diverse policies. Given suitable data, these approaches can monitor many small populations in parallel, where socioeconomic agency and causality are more apparent. They naturally generate daily public forecasts and guidance of adaptive behavior to keep outbreaks under control while preserving select socioeconomic activities.