Integrating Mathematical Models of Behavior and Infectious Disease: Applications to Outbreak Dynamics and Control

Abstract

This dissertation applies mathematical models from several disciplines to represent the complex interaction between human behavior and infectious disease transmission. We first use game theory and evolutionary dynamics to capture contact and risk behaviors such that they can be coupled to compartmental transmission models. We use this behavior-disease approach to model adaptive prophylaxis use during sexually transmitted infeciton (STI) outbreaks as well as burial practices during the 2014 Ebola outbreak using this framework. We then link rational decision theory and quantitative microbial risk assessment (QMRA) to investigate incomplete compliance with household water treatment (HWT) due to usability/efficacy trade-offs. Our analyses of these systems address theoretical and practical consequences of relaxing simplifying assumptions about behavior. We find that feedback between adaptive behavior and disease prevalence in our STI model lead to a range of dynamic outcomes including damped and sustained oscillations as well as complicating the interpretation of the basic reproductive number. For Ebola, our behavior-disease model gives both a superior fit to surveillance data and more accurate final-size forecasts than a reduced model without behavior change. In addition, our model predicts a shift to sanitary burial practices that corresponds to a change in the estimated reporting rate/population at risk and appears to have reduced the force of infection in the later phases of the outbreak. Finally, simulations and optimal risk-reduction solutions from our decision theoretic QMRA model suggest that current efficacy-focused HWT recommendations may be less effective at reducing the burden of diarrheal disease than interventions that prioritize usability and acceptance by the target population. Together, these results demonstrate the potential to improve infectious disease surveillance and control by modeling human behavioral factors that are often simplified or omitted. Such elements can explain the temporal patterns of outbreaks on short and long term scales while behavioral modeling can identify feedbacks that could be exploited to improve the uptake and sustainability of intervention policies.