Modeling adaptive forward-looking behavior in epidemics on networks

The course of an epidemic can be drastically altered by changes in human behavior. Incorporating the dynamics of individual decision-making during an outbreak represents a key challenge of epidemiology, faced by several modeling approaches siloed by different disciplines. Here, we propose an epi-economic model including adaptive, forward-looking behavioral response on a heterogeneous networked substrate, where individuals tune their social activity based on future health expectations. Under basic assumptions, we show that it is possible to derive an analytical expression of the optimal value of the social activity that matches the traditional assumptions of classic epidemic models. Through numerical simulations, we contrast the settings of global awareness -- individuals only know the prevalence of the disease in the population -- with local awareness, where individuals explicitly know which of their contacts are infected. We show that behavior change can flatten the epidemic curve by lowering the peak prevalence, but local awareness is much more effective in curbing the disease early with respect to global awareness. Our work bridges classical epidemic modeling with the epi-economic approach, and sheds light on the effects of heterogeneous behavioral responses in curbing the epidemic spread.