159 - When and where is dengue predictable? Climate extremes and outbreak dynamics across spatial scales, with evidence from Costa Rica

Dengue is a climate-sensitive disease whose transmission dynamics emerge from nonlinear interactions among environmental and epidemiological processes. The extent to which extreme dengue outbreaks are predictable, and under what conditions, remains an open question. In this talk, we present a data-driven framework for evaluating dengue predictability across spatial and temporal scales using Gaussian Process Empirical Dynamic Modeling (GP-EDM) applied to long-term surveillance and climate data from Costa Rica. Results show that predictability varies substantially across space, with local dynamics often heterogeneous and noisy, but that pooling information across regions reveals shared structure. We find that extreme outbreaks are more predictable than endemic periods, and predictability is highest when outbreaks are linked to climate, particularly when temperature and precipitation act jointly. This work demonstrates that dengue predictability is state-dependent, emerging most clearly under structured environmental forcing and highlighting climate-linked extremes as a key target for forecasting and intervention efforts.