DeepDynaForecast: Phylogenetic-informed graph deep learning for epidemic transmission dynamic prediction

Abstract

In the midst of an outbreak or sustained epidemic, reliable prediction of transmission risks and patterns of spread is critical to inform public health programs. Projections of growth or decline among specific risk groups can aid in optimizing interventions, particularly when resources are limited. Phylogenetic trees have been widely used in the detection of transmission chains and high-risk populations. Moreover, tree topology and the incorporation of population parameters (phylodynamics) can be useful to reconstruct the evolutionary dynamics of an epidemic across space and time among individuals. We now demonstrate the utility of phylodynamic trees for infection forecasting in addition to backtracking, developing a phylogeny-based deep learning system, calledDeepDynaForecast. Our approach leverages a primal-dual graph learning structure with shortcut multi-layer aggregation, and it is suited for the early identification and prediction of transmission dynamics in emerging high-risk groups. We demonstrate the accuracy ofDeepDynaForecastusing simulated outbreak data and the utility of the learned model using empirical, large-scale data from the human immunodeficiency virus epidemic in Florida between 2012 and 2020. Our framework is available as open-source software (MIT license) at:https://github.com/lab-smile/DeepDynaForcast.Author SummaryDuring an outbreak or sustained epidemic, accurate prediction of patterns in transmission risk can reliably inform public health strategies. Projections indicating growth or decline of transmission for specific risk groups can significantly enhance the optimization of interventions, especially when resources are limited. To address this, we presentDeepDynaForecast, a cutting-edge deep learning algorithm designed for forecasting pathogen transmission dynamics. Uniquely,DeepDynaForecastwas trained on in-depth simulation data and used more information from the phylogenetic tree of pathogen sequence data than any other algorithm in the field to date, allowing classification of samples according to their dynamics (growth, static, or decline) with incredible accuracy. We evaluated the model’s performance using both simulated outbreak data and empirical, large-scale data from the HIV epidemic in Florida between 2012 and 2020. We concludeDeepDynaForecastrepresents a significant advancement in genomics-mediated pathogen transmission characterization and has the potential to catalyze new research directions within virology, molecular biology, and public health.