Predictive modelling of threshold exceedance and airborne concentrations of Alternaria and Epicoccum spores across bioclimatic regions in Central Europe

Abstract Several airborne fungal spores, such as Alternaria and Epicoccum, are known for their allergenic potential, yet accurately predicting their atmospheric concentrations remains a challenge. This study presents predictive models for estimating daily concentrations and clinically relevant threshold exceedance events of Alternaria and Epicoccum spores, using long-term aerobiological and meteorological data from five cities in Central Europe. Key meteorological predictors, including time-lagged variables, were identified for each location, and interpretable lasso linear and lasso logistic regression models were developed to forecast spore levels up to seven days in advance. The lasso logistic models achieved high accuracy in threshold exceedance predictions, with F1 scores reaching up to 88.6% for Epicoccum. While lasso linear models effectively captured seasonal patterns and timing, they tended to underestimate peak concentrations, particularly for Alternaria, likely due to the sporadic nature of spore release events. Notably, this is the first predictive model developed for Epicoccum, underscoring the need for clinical validation of allergological thresholds. Regional variability in model performance highlights the importance of local calibration and sustained aerobiological monitoring. These models offer a promising foundation for operational spore forecasting systems, supporting both public health advisories and agricultural decision-making.