The highly dynamic nature of infectious diseases poses inherent difficulties in managing them. Some of the greatest concerns about infectious diseases are about what will happen next. When will the flu season peak and will this year's flu season be bigger than last year's? Are a few cases of dengue a sign that a large outbreak is about to happen? How do we make forecasts about a newly emerging disease that we know little about?
The field of temporal forecasting of infectious diseases is experiencing a renaissance, with methods increasing in sophistication all the time. Growing emphasis is being placed on Bayesian techniques, hierarchical models, data assimilation, and ensemble methods.
Infectious disease prevalence is not only dynamic but also highly variable in space. On the one hand, spatial variation in drivers of disease transmission, such as weather and human demography, play an important role in driving this variability. On the other hand, spatial variation in nonlinear feedbacks from population immunity also contribute to this variability. Other factors drive spatial variation not in transmission, per se, but rather in the appearance of disease. For example, the rates at which diseases are reported in different areas vary greatly and spatial aggregation routinely obscures relationships between spatial patterns of disease and their underlying drivers.
Promising methods are being developed at the CICS to deal with these issues that leverage high-resolution geographic information systems, computational modeling, and Bayesian models equipped to handle multiple data types.
Intervention Impact Projections
Model-based projections of location-specific expected numbers of Zika virus infections among childbearing women.
Public health decision-makers are continually faced with hard questions about when, where, to whom, and in what quantity to allocate interventions, including vaccines, drugs, and, in the case of vector-borne diseases, entomological control measures. Sometimes these questions revolve around long-standing interventions, such as vaccines for yellow fever and cholera, whereas other times these questions are asked in the context of newly developed interventions without precedent, such as recently developed vaccines for malaria and dengue. Models used to address these questions tend to involve rich biological detail and range in mathematical forms from differential equations to agent-based simulations.
Coupling model projections of epidemiological impact with models of economic valuation are increasingly needed to inform policy given limited resources and the imperfect nature of interventions to control and prevention of infectious diseases.