The daily pollen forecast provides crucial information for allergic patients to avoid exposure to specific pollens. Pollen counts are typically measured with air samplers and analyzed with microscopy by trained experts. Automated analyses of pollen extracts are being explored as an alternative to traditional pollen counting. METHODS: Extracts of ambient air-sampled pollen from Munich in 2016 and 2017 were lyophilized, rehydrated in optimal NMR buffers, and filtered to remove proteins. NMR spectra were analyzed for pollen associated metabolites. Regression and classification models, using traditional machine learning and deep learning algorithms, were trained to recognize patterns in the metabolites or NMR spectra, based on expertidentified pollen counts. RESULTS: Regression and decision-tree based algorithms using the concentration of metabolites, measured from the NMR spectra, outperformed using the NMR spectra themselves as input data for pollen identification. Categorical prediction algorithms trained for low, medium, high, and very high pollen count groups had accuracies of 74% for the tree, 82% for the grass, and 93% for the weed pollen count. Deep learning models performed better than regression models for NMR spectral input, and were the overall best method in terms of relative error and classification accuracy (86% for tree, 89% for grass, and 93% for weed pollen count). CONCLUSIONS: This study demonstrates that NMR spectra of airsampled pollen extracts could be used in an automated fashion to provide genus and type-specific measures of the pollen count. The classification algorithms can accurately differentiate the low/medium/high category standards of the National Allergy Board.

A clear distinction between the morphology of allergenic pollen grains of various genera of the Poaceae family is an important task in determining the causal allergenic factors in the population. It allows significant improvement of the efficiency of seasonal allergy diagnostics caused by grass pollen. Moreover, it let to perform better predictions of allergenic risks for people, suffering from pollinosis caused by Poaceae pollen. Therefore, the aim of our study was to establish the morphological difference between the pollen grains of plants of various species of Poaceae family in order to further determination of the possibility to use the established distinctions for the identification of pollen in aerobiological studies. For this, both herbarium samples and pollen of the studied plants were collected in the field during May-June 2019 in Vinnytsia. The pollen was shaken off the anthers directly onto a glass slide, immediately stained with basic fuchsin, and covered with a cover slip. The sizes of pollen grains – their width and length – were determined and analyzed using the PhotoM 1.21 program, and the obtained data on the sizes of pollen were divided into categories by the quartile method in Excel. Three categories of pollen sizes were identified: large, medium and small. Large grains had width and length parameters of 40 μm or more, average grains ranged from 26 to 39 μm, and small grains had a size of 26 μm or lesser in width and length. The large category includes the pollen of Hordeum morinum (39.5-53.1 μm), Elytrigia repens (41-48 μm), Secale cereale (48.4-62.5 μm) and Bromus arvensis (42.2-52.7 μm). The medium grain category included pollen from Dactylis glomerata (29.2-38.1 μm), Poa spp. (26.1-37.3 μm), Panicum capillare (33.3-39.5 μm), Lolium perenne (30.4-35.3 μm), Bromus sterilis (28.3-30.8 μm). The pollen size of B. ramosus ranged from 26.1 to 39.5 µm, and B. tectorum was from 35 to 38.4 µm. The pollen grains of Poa pratense (22.1-25.9 μm) and Piptatherum spp were assigned to the category of the smallest pollen (20.3-24.1 microns). Agrostis gigantea was the only grass pollen type whose size fitted for each category. We found out large, medium-sized and grains of 25.0-27.7 microns, which lie between categories 2 and 3, for different populations of this plant. Consequently, some genera and species of Poaceae can be distinguished by the size of their pollen, while in others the size of pollen grains varies considerably. It is necessary to carry out further research that will help to establish the morphology of pollen of a larger number of Grass family plants. This will significantly improve the diagnosis and prevention of seasonal allergy caused by grass pollen in Ukraine.