🤖 AI Summary
In developing epidemic surveillance systems, daily reported data on infections, hospitalizations, deaths, and recoveries often suffer from missingness, inconsistency, and reporting delays. To address this, we propose a dynamic Bayesian statistical modeling framework. Methodologically: (1) we explicitly model the missing-data mechanism, disentangling reporting bias from the underlying epidemiological process; (2) we integrate calendar effects and domain-informed epidemiological priors to enable structured incorporation of expert knowledge; and (3) we build an updateable system for real-time estimation and short-term forecasting. Empirical evaluation on multi-source French COVID-19 data demonstrates substantial improvements in real-time estimation accuracy of key indicators—including the effective reproduction number, hospital burden, and mortality risk—as well as 7–14-day forecast reliability. This work establishes the first reusable, interpretable, and robust statistical modeling benchmark for public health response.
📝 Abstract
This paper addresses statistical modelling and forecasting of key indicators describing the severity of a developing pandemic, using routinely reported daily counts of infections, hospitalizations, deaths (both in and out of hospital), and recoveries. These observed counts constitute what we term ``available data''. Because such data are typically incomplete or inconsistently reported, we address several novel missing data challenges arising in this context and propose statistically rigorous solutions that enable inference based solely on the available information. The model is formulated dynamically, explicitly incorporating calendar effects to capture systematic temporal variations in the progression of the pandemic. The proposed framework is illustrated using data from France collected during the COVID-19 pandemic. Our approach also establishes a new benchmark for integrating prior information from domain experts directly into the modelling process, thereby enabling a potential new division of labour between statistical estimation and epidemiological knowledge from external experts.
