Principal Investigator: Joshua Epstein, PhD
Research Question:
Two major lines of work are proposed, continuing activities well underway in PACER I. They are Behavioral Factors in Crisis Preparedness and Response and the continued collaboration with Dr. Ron Brookmeyer on the development of real-time statistical tools for estimating disease parameters for novel pathogens.
Behavioral Factors in Crisis Preparedness and Response.
We will continue factoring behavior (fear, distrust, biased risk appraisal, emotional contagion) into the modeling of disaster preparedness and containment planning for catastrophic events. Containment of infectious diseases (Pandemic Flu, BT Smallpox), noninfectious agents (Anthrax), airborne chemicals, and radiological events all require compliance with mitigation measures. Yet, we know that high levels of vaccine refusal, flight (producing congestion and exposure), and other forms of noncompliance are likely to degrade mitigation efforts and can exacerbate disasters. These behavioral factors have been virtually ignored in the modeling of infectious diseases, (including pandemic flu and bioterror smallpox) and other major health threats. PACER is seeking to address this important shortcoming by building realistic human behaviors--under novel and extreme stress---into our modeling.
As one example, the PACER research, Epstein et al, "Coupled Contagion Dynamics of Fear and Disease: Mathematical and Computational Explorations" was published in PLoS_1, 3(12), 20081. It models how an epidemic of fear can co-evolve with an epidemic of disease. Using both nonlinear dynamical systems and agent-based computation, we model two interacting contagion processes: one of disease and one of fear of the disease. Individuals can ‘‘contract’’ fear through contact with individuals who are infected with the disease (the sick), infected with fear only (the scared), and infected with both fear and disease (the sick and scared). Scared individuals–whether sick or not–may remove themselves from circulation with some probability, which affects the contact dynamic, and thus the disease epidemic proper. If we allow individuals to recover from fear and return to circulation, the coupled dynamics become quite rich, and can include multiple waves of infection, as occurred in the 1918 Pandemic Flu.
We also study flight as a fear-inspired behavioral response. In a spatially extended setting, even relatively small levels of fear-inspired flight can have a dramatic impact on spatio-temporal epidemic dynamics. Self-isolation and spatial flight are only two of many possible reactions that fear-infected individuals may take. Our main point is that behavioral adaptations must be incorporated into disaster modeling, and we will continue to do that. It should be noted that Behavioral Epidemiology is the focus of the PI Epstein’s 2008 NIH Director’s Pioneer Award. Synergies will therefore benefit both efforts.
Though informed by history, the behavioral model published in PLoS_1 is theoretical, and unfolds on a small-scale abstract lattice. In PACER II, we will scale this analysis up, to investigate the implications of behavioral adaptation in a realistic epidemic on the national scale—particularly as it bears on Surge Capacity. This incorporation of behavioral adaptation into PACER modeling is a primary deliverable. However, deeper understanding of human behavior under circumstances of novel and extreme stress will inform and cut across the other major PACER efforts, including real-time estimation of disease parameters for optimal quarantine, discussed next.
Continue collaboration with Drs Ron Brookmeyer, Nicholas Reich, and Derek Cummings on the development of real-time statistical tools for estimating disease parameters.
In the event of a large-scale outbreak of a novel infectious agent (naturally occurring or engineered), efficiently using available preliminary surveillance information to characterize the new pathogen will be of great value: Good data will be scarce, and rapid and accurate characterization of the incubation period could prove a critical tool for setting public policy.
The incubation period--the length of time between the exposure to an infectious agent and the onset of clinical symptoms—is a key factor for designing effective quarantine policy. Knowing how long individuals can harbor a pathogen and possibly be infectious while not displaying any symptoms is vital to setting an effective quarantine length. If even just a few individuals are infectious but asymptomatic, lifting quarantine before these individuals know they are sick and therefore having them circulate freely in the population could lead to a second wave of epidemic. But since quarantine involve real costs (economic and social), it is important that they do not last longer than necessary. Therefore, especially accurate estimates are needed for the upper tail (e.g. the 95th, 99th percentiles) of the incubation period distribution. PACER-funded research in this field has already received scientific acclaim.
Brookmeyer, Cummings, and Reich are developing new statistical methods for such estimation, but novel pathogens are by definition still unknown. To solve this problem, Hammond and Epstein are creating “virtual” epidemics using agent-based computational modeling. In these models, the incubation period of a “new” pathogen is known, and plausible “questionnaire” data can be collected from cyber-individuals who become sick as the epidemic spreads through an artificial population. These models and the data they generate can be used for three purposes: (1) to illustrate the importance of setting quarantine policy properly (2) to help Brookmeyer et al to hone statistical tools for estimation of incubation periods with realistically imperfect data (3) to help redesign the type of data collected from the sick during an epidemic, so that new and better questionnaires could be deployed in advance of a true pandemic. A preliminary agent-based framework has been implemented, and design of a data protocol is underway.