Valley fever is a fungal disease that lives in the dry soils of the southwestern United States, especially in Arizona. People can get sick when they breathe in the fungus’ spores, which become airborne when the soil is disturbed. Because early symptoms of Valley fever are similar to other illnesses like the flu or pneumonia, it’s often misdiagnosed, leading to delays in proper treatment.

In this study, we aimed to predict future Valley fever cases in Arizona. We used nearly three decades of reported case data (1997–2023), along with information about local weather, soil, and air quality. These environmental conditions are known to influence the growth and spread of the fungus. Our prediction model uses a machine learning technique called XGBoost, and we trained it using county-level data from sources such as NASA and Arizona public health records.

The model successfully captured both seasonal and annual changes in case numbers and performed well when tested on new data from 2024. With this early warning system, we can alert people when the risk of infection is high, especially those who work outdoors and are more likely to come into contact with dust carrying the spores.