Harnessing Computational Tools and Complex Biological Data for Environmental and Health Applications

West Nile Virus (WNV) remains a persistent public health threat in North America, necessitating accurate forecasting models to guide disease surveillance and vector control. This dissertation integrates genomic, epidemiological, meteorological, ecological, and demographic data in an effort to better understand the drivers of WNV outbreaks with the ultimate goal of improving WNV predictions. First, a population genetics analysis of Culex tarsalis, a key WNV vector, identified four distinct locally adapted populations shaped by multiple climatic factors—particularly temperature, precipitation, wind patterns, and daylight length. Second, a national-scale machine learning model incorporating these and other climate variables along with land cover, avian diversity, and human demographics was developed to try and predict WNV case numbers across the continental U.S., but this model ultimately exhibited poor performance due to surveillance data imbalance, inconsistency, and poor temporal granularity. To address the temporal granularity issue, a high-resolution regional model for California was developed, leveraging finer-grained surveillance and environmental data. This model, integrating time-series autoregression and iterative updates, significantly outperformed the national model and showed that the same climate factors driving mosquito adaptation were the most consistent predictors on WNV outbreaks. These findings not only highlight the connection between vector species biology and the diseases they transmit, but also emphasize the need for standardized, higher-quality surveillance data. As climate and ecological shifts continue to influence vector species dynamics and disease spread, data-driven models will be essential for guiding proactive public health interventions.