Perm National Research Polytechnic University (PNRPU), in collaboration with the Higher School of Economics (HSE) and Perm State Pharmaceutical Academy (PSPA), has developed Russia’s first program to combat seasonal pollen allergies. This was reported by the PNRPU press service.
“A neural network-based computer model analyzes weather forecasts in real time, while current pollen concentrations are assessed using aeropalinological monitoring data. Based on this information, the algorithm generates a peak concentration forecast for each allergen, making it possible to accurately predict a rise in morbidity and, consequently, the region’s required supply of antihistamines,” said Konstantin Schwartz, a professor at the university.
According to the researchers, the shift toward predictive models of plant pollination that account for current weather conditions is a global trend. Such solutions rely on global pollen monitoring networks, which collect data using specialized traps and analyze airborne allergen concentrations daily. Until now, Russia lacked its own platform built on local data.
The model was trained on ten years of observational data covering nine of the most allergenic plant species: birch, alder, grasses, maple, elm, pine, poplar, nettle, and ragweed. The algorithm was then calibrated using pharmacy supply data. By integrating these datasets, the system was able to identify how weather conditions trigger pollen releases and how these peaks influence demand for specific medications.
“The model gives us not just a seasonal forecast, but a ‘schedule’ of pollen peaks several days in advance. For example, we can see that tomorrow a birch pollen release is expected to be 12% above the average peak. This means that in 2–3 days, demand for antihistamines will increase by 1.4% of that value. By aggregating these forecasts across all allergens for the entire season, we can calculate the total required procurement volume. So, instead of an average of 10,800 packs of Loratadine for the whole period, we need to stock approximately 11,500 packs ahead of the peak,” Schwartz explained.
