Physics-informed deep learning for infectious disease forecasting

Accurate forecasting of contagious illnesses has become increasingly important to public health policymaking, and better prediction could prevent the loss of millions of lives. To better prepare for future pandemics, it is essential to improve forecasting methods and capabilities. In this work, we propose a new infectious disease forecasting model based on physics-informed neural networks (PINNs), an emerging area of scientific machine learning. The proposed PINN model incorporates dynamical systems representations of disease transmission into the loss function, thereby assimilating epidemiological theory and data using neural networks (NNs). Our approach is designed to prevent model overfitting, which often occurs when training deep learning models with observation data alone. In addition, we employ an additional sub-network to account for mobility, vaccination, and other covariates that influence the transmission rate, a key parameter in the compartment model. To demonstrate the capability of the proposed model, we examine the performance of the model using state-level COVID-19 data in California. Our simulation results show that predictions of PINN model on the number of cases, deaths, and hospitalizations are consistent with existing benchmarks. In particular, the PINN model outperforms the basic NN model and naive baseline forecast. We also show that the performance of the PINN model is comparable to a sophisticated Gaussian infection state space with time dependence (GISST) forecasting model that integrates the compartment model with a data observation model and a regression model for inferring parameters in the compartment model. Nonetheless, the PINN model offers a simpler structure and is easier to implement. Our results show that the proposed forecaster could potentially serve as a new computational tool to enhance the current capacity of infectious disease forecasting.

Optimal non-pharmaceutical intervention policy for Covid-19 epidemic via neuroevolution algorithm

National responses to the Covid-19 pandemic varied markedly across countries, from business-as-usual to complete shutdowns. Policies aimed at disrupting the viral transmission cycle and preventing the healthcare system from being overwhelmed, simultaneously exact an economic toll. We developed a intervention policy model that comprised the relative human, economic and healthcare costs of non-pharmaceutical epidemic intervention and arrived at the optimal strategy using the neuroevolution algorithm. The proposed model finds the minimum required reduction in contact rates to maintain the burden on the healthcare system below the maximum capacity. We find that such a policy renders a sharp increase in the control strength at the early stages of the epidemic, followed by a steady increase in the subsequent ten weeks as the epidemic approaches its peak, and finally control strength is gradually decreased as the population moves towards herd immunity. We have also shown how such a model can provide an efficient adaptive intervention policy at different stages of the epidemic without having access to the entire history of its progression in the population. This work emphasizes the importance of imposing intervention measures early and provides insights into adaptive intervention policies to minimize the economic impacts of the epidemic without putting an extra burden on the healthcare system.