Flexible Modeling of Epidemics with an Empirical Bayes Framework

TL;DR

This paper introduces a flexible, data-driven Empirical Bayes framework for real-time epidemic forecasting, capable of modeling various epidemic behaviors and providing complete posterior distributions, demonstrated on influenza data.

Contribution

It presents a novel semiparametric Bayesian approach that captures epidemic variability without strict domain assumptions, improving forecast accuracy and applicability to other diseases.

Findings

Accurately predicted influenza season onset, peak, and duration.

Produced comprehensive posterior distributions for epidemic forecasts.

Outperformed baseline models in cross-validation tests.

Abstract

Seasonal influenza epidemics cause consistent, considerable, widespread loss annually in terms of economic burden, morbidity, and mortality. With access to accurate and reliable forecasts of a current or upcoming influenza epidemic's behavior, policy makers can design and implement more effective countermeasures. We developed a framework for in-season forecasts of epidemics using a semiparametric Empirical Bayes framework, and applied it to predict the weekly percentage of outpatient doctors visits for influenza-like illness, as well as the season onset, duration, peak time, and peak height, with and without additional data from Google Flu Trends, as part of the CDC's 2013--2014 "Predict the Influenza Season Challenge". Previous work on epidemic modeling has focused on developing mechanistic models of disease behavior and applying time series tools to explain historical data. However, these models may not accurately capture the range of possible behaviors that we may see in the future. Our approach instead produces possibilities for the epidemic curve of the season of interest using modified versions of data from previous seasons, allowing for reasonable variations in the timing, pace, and intensity of the seasonal epidemics, as well as noise in observations. Since the framework does not make strict domain-specific assumptions, it can easily be applied to other diseases as well. Another important advantage of this method is that it produces a complete posterior distribution for any desired forecasting target, rather than mere point predictions. We report prospective influenza-like-illness forecasts that were made for the 2013--2014 U.S. influenza season, and compare the framework's cross-validated prediction error on historical data to that of a variety of simpler baseline predictors.