Leyao Zhang

Department of Biostatistics, School of Public Health, University of Michigan

Title: Optimal monitoring of infectious disease and inference using a transitional prediction model

Abstract: The effectiveness of tracking delta virus infected cases has been compromised by at-home COVID test kits. An alternative solution to monitoring contagions of the COVID disease in the population is to survey viral loads from sewage water systems. In a city, numerous sewage manholes create a complex interconnected network of potential sampling sites. Limited resources prevent sampling from every manhole. The central scientific question is to select important manholes that are most relevant to the prediction of confirmed infectious cases in a specific community with suitable uncertainty quantification. In this paper, we develop a supervised learning paradigm of time-series transitional models via the mixed integer optimizer to identify important sampling sites for cost-effective disease monitoring. We establish the theoretical guarantees of the selection consistency for the proposed methodology. A novel self-conformal inference framework is proposed to quantify estimation uncertainty by the means of adversarial noise perturbation. Our methodology is evaluated by extensive simulation studies and illustrated by a motivating data example of a city COVID monitoring program.

The talks will be both in-person at Chapman Hall 435 and on Zoom: https://unc.zoom.us/j/96417451559.