Accounting for a typology of travel behavior in modeling infectious disease transmission, detection, and community response

Modifying travel behavior is often critical for the public health response to infectious disease outbreaks, but depends on consistent, reliable detection. Often, this involves recommending that communities avoid high-risk areas, but if detection is low where transmission is high, these recommendations could lead to a counterproductive response if cases reported by health systems are mostly from low-risk areas. Mismatches between detection and transmission can be further exacerbated by behavioral associations between locations. For example, locations associated with high vectorborne disease (VBD) risk, such as green spaces and parks, are often distant from doctor’s offices, meaning that more time spent in areas with high VBD risk could mean decreased propensity to visit a doctor when sick. These kinds of behavioral correlations between high-risk areas and doctor’s offices carry especially strong risk of causing a counterproductive community response. We combine mathematical models with mobility data analysis to explore how correlations between trips to high-risk areas and healthcare providers can shape disease detection, behavioral response, and outbreak dynamics. Through modeling, we find that a negative correlation between high-risk places and healthcare facilities can cause community response to worsen the outbreak, due to many undetected cases in high-risk areas. We then use smartphone location data to show that there is a clear dichotomy in terms of travel behavior, where the number of trips to places typically associated with urban areas are strongly positively correlated through space and time, and are strongly negatively correlated with places associated with rural areas. This suggests a simple delineation (urban vs. rural) that can be used in future studies to further elucidate how changes in travel behavior could impact the spread and control of infectious diseases.