TITLE:  When to Respond: A Multi-Agent Stochastic Alert Threshold Model for Declaring a Disease Outbreak

SPEAKER:  Julie Simmons Ivy, Associate Professor, Edward P. Fitts Department of Industrial and Systems Engineering, North Carolina State University 

ABSTRACT:

Influenza pandemics are considered one of the most significant and widely spread threats to public health. In this research, we explore the relationship between local and state health departments with respect to issuing alerts and responding to a potential disease outbreak such as influenza. We modeled the public health system as a multi-agent (or decentralized) partially observable Markov decision process where local and state health departments are decision makers. The model is used to determine when local and state decision makers should issue an alert or initiate mitigation actions such as vaccination in response to the existence of a disease threat. The model incorporates the fact that health departments have imperfect information about the exact number of infected people. The objective of the model is to minimize both false alerts and late alerts while identifying the optimal timing for alerting decisions. Providing such a balance between false and late alerts has the potential to increase the credibility and efficiency of the public health system while improving immediate response and care in the event of a public health emergency. Using data from the 2009-2010 H1N1 influenza outbreak to estimate model parameters including observations and transition probabilities, computational results for near optimal solutions are obtained.  In order to gain insight regarding the structure of optimal policies at the local and state levels, various model parameters including false and late alerting costs are explored. 

This research is a part of the North Carolina Preparedness and Emergency Response Research Center (NCPERRC) and was supported by the Centers for Disease Control and Prevention (CDC) Grant 1PO1 TP 000296-02.