Primary biological aerosols are released into the air both by human activity and by natural processes from their sources (i.e. plants, fungi, and other biota). Many bioaerosols contain allergens, which affect human health. while others are cause of plant diseases thus being of importance for forestry and agriculture. Finally, there is growing evidence that bioaerosols have a role in cloud microphysics including ice nucleation processes.
Biological component of airborne particulate matter is commonly assessed by manual methods which ignores some of the important components and limits the temporal resolution. Therefore, emerging new measurement methods based on high-end imaging and laser induced fluorescence are expected to meet end-user needs by provide automatic and real-time data flow. However, particle type recognition remains a specific challenge.
The aim of our research is to couple big data collected by new generation automatic single particle analyser (Rapid-E) and advanced artificial intelligence tools to enable identification and quantification specific bioaerosols in outdoor air.
The tests of developed system in operational monitoring proved that laser spectroscopy-based automatic bioaerosol measurements equipped with our machine learning algorithms enable the identification and quantification in high temporal resolution a full variety of airborne particles of biological origin. We have successfully quantified 16 the most abundant and allergy relevant airborne pollen types. High temporal resolution measurements (up to sub-minute) revealed 15% better insight into explanation of allergy total symptoms for selected individuals in the study area. In addition to airborne pollen, system detected short but very intensive episodes of airborne starch particles. Thanks to hight temporal resolution of bioaerosol measurements, wind direction analysis and dispersion modelling grain storage towers were identified as sources.
The new generation of automatic devices coupled with advanced analytical tools and extensive reference datasets for training bioaerosol classification models have the potential to increase our understanding of the distribution of bioaerosols in the air, provide insights into biological processes such as release and dispersal mechanisms, and allow us to conduct investigations into dose-response relationships and personal exposure to aeroallergens and plant pathogens.