INBRE Undergrad researcher in Coeur d’Alene pioneers AI use in predictive modeling for Valley Fever spread

Undergrad researcher in Coeur d’Alene pioneers AI use in predictive modeling for Valley Fever spread

University of Idaho Coeur d’Alene computer science student Leif Huender proves that real-world discovery isn’t reserved for seasoned scholars.

Huender’s undergraduate research reveals how airborne fungal diseases spread through weather patterns while pioneering the innovative use of artificial intelligence in outbreak prediction.

“Our research is the first to apply extended LSTM models to Valley Fever prediction. This opens new doors for public health planning and disease prevention.” Leif Huender, computer science undergraduate researcher

As wildfires intensify worldwide, researchers are exploring the behaviors and health effects of fires in new ways. Huender — a first-generation college student who is concurrently enrolled at U of I Coeur d’Alene and North Idaho College — is part of a team developing AI-driven methods to predict disease outbreaks from wildfire smoke using long short-term memory (LSTM) models.

LSTM models are a type of artificial intelligence that can remember important information over time and forget irrelevant details, making them good at predicting what comes next in sequences, like in text or weather patterns.

“Using LSTM models, we’ve made a big step forward in predicting future outcomes,” Huender said. “These models are exceptional at recognizing complex patterns over time, even with limited data.”

Predicting Valley Fever

Valley Fever, a fungal disease common in the arid regions of California and Arizona, has long been linked to weather patterns, but predicting outbreaks has remained a challenge.

Huender’s undergraduate research explores alternative ways to forecast Valley Fever outbreaks using a new approach with AI.

With an INBRE fellowship working with Center for Intelligent Industrial Robotics Associate Director and Vandal alumna Mary Everett, Huender developed AI methods predicting agricultural outcomes in microclimates with limited weather data. He presented this work at the International Society of Precision Agriculture in July 2024.

“I realized the same techniques we used to predict crop yields could be applied to forecasting disease outbreaks,” Huender said. “Both problems involve analyzing complex environmental data to make real-world predictions. The LSTM models that worked for agriculture proved equally valuable for forecasting Valley Fever outbreaks based on climate conditions.”