Analysis and Methods to Mitigate Effects of Under-reporting in Count Data

Abstract

Under-reporting of count data poses a major roadblock for prediction and inference. In this paper, we focus on the Pogit model, which deconvolves the generating Poisson process from the censuring process controlling under-reporting using a generalized linear modeling framework. We highlight the limitations of the Pogit model and address them by adding constraints to the estimation framework. We also develop uncertainty quantification techniques that are robust to model mis-specification. Our approach is evaluated using synthetic data and applied to real healthcare datasets, where we treat in-patient data as 'reported' counts and use held-out total injuries to validate the results. The methods make it possible to separate the Poisson process from the under-reporting process, given sufficient expert information. Codes to implement the approach are available via an open source Python package.

Predicting COVID-19 community infection relative risk with a Dynamic Bayesian Network

“…There has been recent research emphasis placed on modeling the effect of under-reporting of COVID-19 and other infectious diseases ( 128 – 130 ). We did not account for this in our modeling due to some of the intrinsic limitations in INLA (the package in R used for the Bayesian hierarchical modeling); lacking the ability to directly account for this ( 47 ).…”