We propose, implement, and evaluate a method to estimate the daily number of new symptomatic COVID-19 infections, at the level of individual U.S. counties, by deconvolving daily reported COVID-19 case counts using an estimated symptom-onset-to-case-report delay distribution. Importantly, we focus on estimating infections in real-time (rather than retrospectively), which poses numerous challenges. To address these, we develop new methodology for both the distribution estimation and deconvolution steps, and we employ a sensor fusion layer (which fuses together predictions from models that are trained to track infections based on auxiliary surveillance streams) in order to improve accuracy and stability.
MJ was supported by a fellowship from the Center for Machine Learning and Health at Carnegie Mellon.
AC and RJT were supported by a gift from Google.org.
The authors are grateful to Logan Brooks, Roni Rosenfeld, James Sharpnack, Sam Abbott, Joel Hellewell, and Sebastian Funk for several early insightful conversations.
"Real-Time Estimation of COVID-19 Infections: Deconvolution and Sensor Fusion." Statist. Sci. 37 (2) 207 - 228, May 2022. https://doi.org/10.1214/22-STS856