The goal of this project is to make transformative progress in the simulation of infectious diseases. The recent COVID-19 pandemic demonstrated the importance and potential of epidemiological models. It also revealed short-comings of the employed methods which resulted in limited predictive power and ambiguity concerning the effectiveness of control strategies. Besides uncertainty in the available data, this can be attributed to approximations that largely ignore the heterogeneity of the population.
In this project, we want to develop and demonstrate a novel technique to enable a more realistic and practically useful analysis of infectious-disease dynamics that captures the heterogeneity in the population with respect to behavioral patterns and bio-medically relevant parameters of individuals, their response to information and health policy changes, as well as the stochastic nature of infection processes. The specific goals are the following.
(1) We will implement a new method for the simulation of stochastic network dynamics, inspired by numerical techniques in quantum many-body physics. In contrast to Markov-chain Monte Carlo, it samples entire trajectories of the evolution and we can, thus, study dynamics conditioned on events like a big epidemic outbreak.
(2) The ability to study the dynamics conditioned on certain (possibly rare) events, enables an efficient risk-factor analysis. We will demonstrate this by analyzing the impact of the agents’ behavioral patterns and relevant bio-medical parameters. This concerns, for example, the frequency and duration of social interactions, gathering sizes, as well as the variability of pathogen transmissibility and morbidity in the population. Furthermore, we can assess and optimize prevention and mitigation measures like lockdowns, social distancing, vaccination etc.
(3) The aforementioned agent parameters vary widely in the population. In collaboration with sociologists and global-health experts we will identify suitable real-world data for the distribution of these parameters and incorporate it in our models. The risk-factor analysis may also identify important gaps in the available data, e.g., concerning correlations between different parameters.
(4) Depending on the progress on goals 1-3, we will begin to investigate the impact of the agents’ dynamical response to health policy changes and information that they gather during the course of an epidemic. This can, for example, be information about infection rates, hospitalizations, and deaths as gained through local interactions or, globally, through news media.