Forecasting COVID-19 Counts At A Single Hospital: A Hierarchical Bayesian Approach

TL;DR

This paper introduces hierarchical Bayesian models for accurately forecasting daily COVID-19 hospitalizations at individual hospitals, leveraging count data, temporal dependencies, and shared information across sites.

Contribution

The paper presents novel hierarchical Bayesian models tailored for hospital-specific COVID-19 count forecasting, incorporating count data, temporal dynamics, and cross-site information sharing.

Findings

Models outperform baseline methods in forecasting accuracy.

Approach generalizes across hospitals in different countries.

Prospective evaluation shows improved 2-week-ahead predictions.

Abstract

We consider the problem of forecasting the daily number of hospitalized COVID-19 patients at a single hospital site, in order to help administrators with logistics and planning. We develop several candidate hierarchical Bayesian models which directly capture the count nature of data via a generalized Poisson likelihood, model time-series dependencies via autoregressive and Gaussian process latent processes, and share statistical strength across related sites. We demonstrate our approach on public datasets for 8 hospitals in Massachusetts, U.S.A. and 10 hospitals in the United Kingdom. Further prospective evaluation compares our approach favorably to baselines currently used by stakeholders at 3 related hospitals to forecast 2-week-ahead demand by rescaling state-level forecasts.