P-247 - A Systematic Review of Predictive Models for Fusarium Head Blight in Wheat

Fusarium head blight (FHB) poses a significant threat to global wheat production, leading to yield losses and the contamination of grain with deoxynivalenol (DON), raising concerns about food safety. Accurate forecasts are essential for effective management, resulting in the development of various forecasting models. This systematic review assesses prediction models for FHB and DON, highlighting methodologies, performance metrics, and existing research gaps. We searched four electronic databases for the keywords ‘wheat’ and ‘FHB’ using various terms without applying limits or filters. We screened the titles and abstracts of 14,258 papers and conducted full-text reviews on 406 articles specifically focused on FHB or DON forecasting. Preliminary findings indicate that weather variables are the most frequently used predictors, with temperature and humidity being the key factors. Some models also incorporate rainfall and the crop growth stage to enhance accuracy. The models can be classified into three categories: statistical (e.g., logistic regression), machine learning (e.g., neural networks, random forests), and mechanistic (e.g., weather-based risk models). Many studies rely on internal validation, limiting their applicability across different agroecosystems. This review emphasizes the necessity for external validation, the use of real-time sensor data, and the integration of genetic factors to improve forecasting. Future research should focus on ensemble modeling and decision support systems to strengthen FHB risk prediction and disease management for wheat growers.