Estimating the State of Epidemics Spreading with Graph Neural Networks

Abstract

When an epidemic spreads into a population, it is often unpractical or impossible to have a continuous monitoring of all subjects involved. As an alternative, algorithmic solutions can be used to infer the state of the whole population from a limited amount of measures. We analyze the capability of deep neutral networks to solve this challenging task. Our proposed architecture is based on Graph Convolutional Neural Networks. As such it can reason on the effect of the underlying social network structure, which is recognized as the main component in the spreading of an epidemic. We test the proposed architecture with two scenarios modeled on the CoVid-19 pandemic: a generic homogeneous population, and a toy model of Boston metropolitan area.

Epidemic inference through generative neural networks

“…Neural networks have already been applied to epidemic forecasting [41][42][43] but rarely to epidemic inference and reconstruction problems. Two recent preliminary works apply neural networks to epidemic inference problems: in [44] the patient zero problem is tackled using graph neural networks, while a similar technique is applied to epidemic risk assessment in [45]. The presented approach allows to address successfully a large class of epidemic inference problems, ranging from the patient-zero problem and individual risk assessment to the inference of the parameters of the propagation model under a unique neural network framework.…”

Effective vaccination strategy using graph neural network ansatz

“…GNN has become a widely used method for network analysis because of its convincing performance in various fields, such as estimation of molecular properties [30,31], drug discovery [32], and traffic forecasting [33,34]. In the epidemic field, GNNs have been employed for the prediction of disease prevalence [35][36][37], identification of patient zero [38], and estimation of epidemic state using limited information [39]. Few studies have developed dynamic epidemic control schemes that identify epidemic hotspots from partially observed epidemic state of each individual [40,41].…”