Estimating Time-Varying Epidemic Severity Rates with Adaptive Deconvolution

2025-10-17

AI Summary

In public health, time-varying epidemic severity measures—such as case fatality ratios—are confounded by viral evolution, interventions, and reporting delays, rendering conventional ratio estimators substantially biased. To address this, we propose an adaptive deconvolution framework regularized by trend filtering, enabling robust extraction of dynamic risk signals from delayed and aggregated count data. Grounded in a Poisson–binomial statistical model, our method employs an approximate maximum likelihood solution to yield locally adaptive, smooth estimates of time-varying severity. It effectively corrects for both reporting delays and aggregation bias, outperforming standard ratio-based approaches on both real-world and simulated COVID-19 data, while exhibiting robustness to model misspecification. Our key contribution is the first integration of trend filtering into deconvolution-based estimation, yielding a dynamic severity modeling framework that simultaneously ensures theoretical interpretability, computational tractability, and suitability for near-real-time analysis.

Technology Category

• Reasoning under Uncertainty: Graphical Models
• Search and Optimization: Non-convex Optimization
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Several key metrics in public health convey the probability that a primary event will lead to a more serious secondary event in the future. These "severity rates" can change over the course of an epidemic in response to shifting conditions like new therapeutics, variants, or public health interventions. In practice, time-varying parameters such as the case-fatality rate are typically estimated from aggregate count data. Prior work has demonstrated that commonly-used ratio-based estimators can be highly biased, motivating the development of new methods. In this paper, we develop an adaptive deconvolution approach based on approximating a Poisson-binomial model for secondary events, and we regularize the maximum likelihood solution in this model with a trend filtering penalty to produce smooth but locally adaptive estimates of severity rates over time. This enables us to compute severity rates both retrospectively and in real time. Experiments based on COVID-19 death and hospitalization data, both real and simulated, demonstrate that our deconvolution estimator is generally more accurate than the standard ratio-based methods, and displays reasonable robustness to model misspecification.

Problem

• Estimating time-varying epidemic severity rates from aggregate count data
• Correcting bias in ratio-based estimators for secondary event probabilities
• Developing adaptive deconvolution method for real-time severity monitoring

Innovation

• Adaptive deconvolution estimates time-varying epidemic severity rates
• Poisson-binomial model with trend filtering regularization
• Real-time retrospective computation using COVID-19 data