Bayesian Hierarchical Reporting Delay Model in Infectious Disease Forecasting

• Activity Number: 355 - Contributed Poster Presentations: Section on Bayesian Statistical Science
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Krzysztof Sakrejda* and Nicholas G. Reich and Stephen Lauer
• Companies: University of Massachusetts - Amherst and University of Massachusetts Amherst and University of Massachusetts - Amherst
• Keywords: Bayesian ; hierarchical ; forecasting ; survival ; infectious disease

Abstract

Time-series of incidence have historically been used to track disease burden in epidemics and more recently applied to infectious disease incidence forecasting. Individual records of disease occurrence, used to construct these time series, are often observed at a significant lag for operational reasons leading to artifacts in time-series and reduced forecast efficiency. We illustrate the issue using records of dengue incidence in Thailand and an ongoing dengue forecasting project. We use a Bayesian hierarchical survival model to represent the delays in case reporting for ~8700 subdistrict in Thailand. We show that using this model to estimate final case counts based on observed case counts improves time-series modeling results and we present a simple integrated time-series and reporting model to carry out the task directly.