D3.484 - Non-Invasive Detection of Airborne Fungal Exposure Using Machine Learning–Assisted Volatile Biomarker Sensing

Indoor fungal growth is a significant contributor to respiratory allergies and asthma exacerbation. However, objective and non-invasive methods for assessing fungal exposure remain limited. Microbial volatile organic compounds (MVOCs), generated during fungal metabolism, have emerged as promising indirect biomarkers for detecting indoor fungal contamination. Reliable real-time detection of these biomarkers under realistic indoor conditions remains a technical challenge, particularly under high humidity.

This study developed a machine learning-assisted multimodal gas sensing platform for the identification of fungal-related volatile biomarkers. Three distinct sensing materials were integrated into different sensor types to produce multidimensional response patterns. Representative MVOCs (ethanol, 2-butanone, and benzene) were evaluated under both single-gas and mixed-gas conditions. In addition, headspace emissions from fungal cultures (Aspergillus, Cladosporium, and Penicillium) were tested under high-humidity environments. Machine learning algorithms, including random forest classification and CatBoost regression, were applied for volatile discrimination and mixed-gas quantification.

Random forest classification achieved 95% accuracy in distinguishing individual volatile compounds. CatBoost regression successfully quantified mixed-gas components, reaching R² values of up to 0.93. The sensing platform differentiated fungal emissions from non-fungal controls with an accuracy of 0.963. Importantly, stable performance was maintained even under saturated humidity conditions.

The proposed multimodal sensing platform, combined with machine learning, demonstrates high sensitivity, selectivity, and robustness for detecting fungal-related volatile biomarkers in complex indoor environments. This approach provides a practical framework for real-time monitoring systems aimed at early detection of indoor fungal growth.