290 - Predicting Geographic Expansion and Genetic Adaptation of Aedes aegypti in Southern Nevada using Machine Learning Techniques

Abstract: Starting in 2017, Aedes aegypti, the main carrier of dengue, Zika, and chikungunya, has quickly expanded across Clark County, Nevada, requiring immediate public health actions. The Southern Nevada Health District has reported Ae. aegypti's presence in around 67 zip codes in 2024. The growing number of dengue cases in nearby states—188 in California and 32 in Arizona in 2023 (confirmed by the CDC)—combined with the identification of dengue-positive Ae. aegypti specimens in Arizona, highlights the rising threat of this invasive species. Unsupervised learning approaches, including clustering algorithms and neural networks, are employed to analyze genomic data, aiming to uncover hidden patterns and reveal new population clusters and genetic markers associated with environmental adaptation. A custom dataset is being developed using Google Earth Engine to incorporate climatic variables for each sample’s location and collection date, enabling precise modeling of the ecological factors driving Ae. aegypti distribution across southern Nevada. Between 2022 and 2023, 52 Ae. aegypti specimens were collected from 18 zip codes, and whole genome sequencing revealed two distinct genetic clusters, suggesting multiple introduction events from genetically diverse populations. Additionally, 86 genes were identified as exhibiting evidence of environmental adaptation, highlighting the mosquitoes’ ability to thrive in the arid conditions of southern Nevada. These findings establish a critical genetic baseline for monitoring the evolutionary dynamics of Ae. aegypti, providing vital insights for improving vector control strategies and mitigating the threat of vector-borne diseases in the southwestern United States. In addressing this public health issue, predictive modeling methods like random forests and gradient boosting are being used to predict the future distribution of Ae. aegypti. These models use genetic markers, environmental factors like temperature, humidity, and topo-climatic variables, along with historical population data.