Policy support for managing the COVID pandemic through artificial intelligence

In recent years, epidemiological modeling has made important progress, and now provides us with a variety of models ranging from the high-level compartmental models, meta-population models, network models to fine-grained individual-based models. Such models allow for simulations which can be combined with advanced optimization approaches using artificial intelligence in order to identify suitable prevention and containment measures. In this project we will extend state-of-the-art Reinforcement Learning techniques, which have been shown to outperform the currently used techniques by epidemiologists to come up with prevention and containment measures. Our approach will take into account different factors of uncertainty, both of the epidemic as well as human factors. Hereby, taking into account different criteria, such as health factors (e.g. hospital load and death counts), but also economic and social impact. We allow for multi-criteria optimization, such that policy makers can trade-off different aspects. We also pay attention to the communication of the outcome of the learning process to the user, by building upon research on explainable reinforcement learning. The research will form the basis for a valuable tool for decision makers when confronted with a pandemic such as COVID-19, even when information on epidemics only gradually becomes available.