Applications of Modeling to Hantavirus, Cholera, and COVID-19

Many rodent-borne hantaviruses are zoonotic pathogens that can cause disease in humans. To evaluate the prevalence of Jaborá virus (JABV) over time within its rodent reservoir, Akodon montensis, we formulated a discrete time model with multiple rodent age classes and unique infection class progression features. We parameterized the model with data collected from a survey of JABV harbored by Akodon montensis in the Mbaracayú Reserve in Paraguay. Our model incorporates three infection types and a recovered class. A new feature of the model allows transition from the latent to the persistently-infected class. With a more complete age and infection structure, we are better able to identify the driving forces of epidemiology of hantaviruses in rodent populations.

Optimal control (OC) applied to epidemic modeling can be a key tool in determining effective intervention strategies for infectious diseases. However, the common formulations of OC problems of infectious disease systems of ordinary differential equations do not always offer features needed by disease managers. These problems often minimize both the disease burden and the intervention cost over a time period to determine the optimal intervention strategy, not allowing for the incorporation of details such as budget constraints. We formulated and analyzed several OC formulations using a basic example of cholera, incorporating features such as a budget constraint or a desired epidemiological goal, to reframe OC problems applied to infectious disease models in a more accessible manner for public health managers.

There is evidence of geographic disparities in COVID-19 hospitalization risks that, if identified, could guide control efforts. Geographic disparities in distribution of socioeconomic, demographic and health-related predictors of ZCTA-level COVID-19 hospitalizations in St. Louis were investigated using choropleth maps. These predictors were then investigated using global negative binomial and local geographically weighted negative binomial models. COVID-19 hospitalization risks were significantly higher in ZCTAs with high diabetes hospitalization and COVID-19 risks, black population, and populations with some college education. The impacts of the first three factors vary by geographical location, implying that a `one-size-fits-all’ approach may not be appropriate for management and control. These findings are useful for informing health planning and guiding vaccination efforts.