Predicting Valley Fever Outbreaks: Novel Mechanistic Models Incorporating Climate and Ecological Interactions.

Coccidioidomycosis (Valley fever) is an environmentally acquired fungal infection endemic to the arid Americas, presenting a growing public health challenge as changing environmental patterns threaten to amplify exposure risks across both established and newly recognized endemic zones. Historically, forecasting efforts have relied on statistical correlations with meteorological variables. These phenomenological models often fail to capture the complex, non-linear interactions between the saprobic (environmental) and parasitic (host) life cycles of Coccidioides, particularly under non-stationary climate conditions. Here, we present a hierarchy of mechanistic Ordinary Differential Equation (ODE) models that explicitly map environmental drivers to the distinct biological stages of the fungal life cycle. We developed successive model iterations, incrementally incorporating soil moisture retention, temperature-dependent growth rates, and wildlife reservoir dynamics, and calibrated them against human case data from various regions of Arizona. We derive a time-variant environmental reproduction number and test how transmission potential fluctuates dynamically with environmental forcing. The comparative forecasting analysis, utilizing various statistical tests, information criteria, Relative Root Mean Square Error, the Diebold-Mariano test, and the Modified Diebold-Mariano, shows how the models progress. Mechanistic models based solely on continuous fungal growth perform worse than statistical baselines. By integrating climate data, we increase predictive power to a level comparable to that of the statistical model. Explicitly incorporating a wildlife reservoir as a biological amplifier significantly improves model forecasting over statistical baselines. This framework offers public health officials a biologically grounded tool to predict disease burden and guide targeted interventions responding to changing climate patterns.