Characterization of partially observed epidemics through Bayesian inference: application to COVID-19
- PDF / 5,446,649 Bytes
- 21 Pages / 595.276 x 790.866 pts Page_size
- 90 Downloads / 156 Views
ORIGINAL PAPER
Characterization of partially observed epidemics through Bayesian inference: application to COVID-19 Cosmin Safta1 · Jaideep Ray1 · Khachik Sargsyan1 Received: 2 July 2020 / Accepted: 29 July 2020 © National Technology and Engineering Solutions of Sandia, LLC 2020
Abstract We demonstrate a Bayesian method for the “real-time” characterization and forecasting of partially observed COVID-19 epidemic. Characterization is the estimation of infection spread parameters using daily counts of symptomatic patients. The method is designed to help guide medical resource allocation in the early epoch of the outbreak. The estimation problem is posed as one of Bayesian inference and solved using a Markov chain Monte Carlo technique. The data used in this study was sourced before the arrival of the second wave of infection in July 2020. The proposed modeling approach, when applied at the country level, generally provides accurate forecasts at the regional, state and country level. The epidemiological model detected the flattening of the curve in California, after public health measures were instituted. The method also detected different disease dynamics when applied to specific regions of New Mexico. Keywords Markov Chain Monte Carlo · Pseudo-marginal MCMC · Bayesian framework · COVID-19 · Infection rate · Incubation model
1 Introduction In this paper, we formulate and describe a data-driven epidemiological model to forecast the short-term evolution of a partially-observed epidemic, with the aim of helping estimate and plan the deployment of medical resources and personnel. It also allows us to forecast, over a short period, the stream of patients seeking medical care, and thus estimate the demand for medical resources. It is meant to be used in the early days of the outbreak, when data and information about the pathogen and its interaction with its host population is scarce. The model is simple and makes few demands on our epidemiological knowledge of the pathogen. The method is cast as one of Bayesian inference of the latent infection rate (number of people infected per day), conditioned on a time-series of For the Special Issue: “Modeling and Simulation of Infectious Diseases-Propagation, decontamination and mitigation”.
B
Cosmin Safta [email protected] Jaideep Ray [email protected] Khachik Sargsyan [email protected]
1
daily new (confirmed) cases of patients exhibiting symptoms and seeking medical care. The model is demonstrated on the COVID-19 pandemic that swept through the US in spring 2020. The model generalizes across a range of host population sizes, and is demonstrated at the country-scale as well as for a sparsely populated desert region in Northwestern New Mexico, USA. Developing a forecasting method that is applicable in the early epoch of a partially-observed outbreak poses some peculiar difficulties. The evolution of an outbreak depends on the characteristics of the pathogen and its interaction with patterns of life (i.e., population mixing) of the host population, both of which are ill-define
Data Loading...
