Understanding the effect of Agricultural Practices on Valley Fever through a Dual -Model Approach using Environmental Factors

Overview

Valley Fever (VF), or coccidioidomycosis, is a fungal infection influenced by environmental factors such as dust exposure. This project investigates the impact of different agricultural practices (APs) on VF epidemiology through a dual -model computational approach. By merging environmental and VF case data from Maricopa County, Arizona, this study employs a Gradient Boosting Regressor (GBR) and a Multilayer Perceptron (MLP) to predict VF case trends and assess the effectiveness of various APs in reducing di sease incidence. The research framework consists of four key stages: (1) data collection and preprocessing of environmental factors such as PM2 emissions, temperature, precipitation, and sunlight, (2) model development and calibration using GBR and MLP to capture both steady trends and extreme fluctuations, (3) experimental validation using an Arduino -based sensor system to measure the effects of APs, and (4) comparative analysis of predicted versus actual VF case reductions. The model achieved a R² score of 0.903, explaining 90.03% of VF case variance, with a mean absolute error (MAE) of 97.197 cases. The results indicate that mulch is the most effective AP, reducing VF cases by 21.32%, compared to 15.39% with organic material cover. Future research will focus on expanding datasets from CDC, NOAA, and NASA, integrating population density metrics, testing alternative models like LSTM and XGBoost, and evaluating additional APs such as salt brine treatments to increase predictive accuracy and real-world applicability.