Application of Machine Learning in Predicting the Impact of Air Pollution on Bacterial Flora

Abstract

Air pollution is recognized by WHO as a cause for 7.6% of global mortality. Ambient air pollution as well as household pollution in the cities rises as global worldwide problem. Aim of the study: The possibility of gaining new knowledge, through decision tree models, of the relationship between the conditions that favors the growth of the bacterial flora related to air pollution factors, like PM2.5 and PM10. Material and methods: predictive cluster trees in CLUS system were obtained with relevant microbiological data form indoor air samples and two locations for outdoor and one indoor samples for PM readings as well as O2, NO2, SO2 and CO. These measurements were performed by two apparatus: BAM-1020 Met One instruments Inc. and Aerocet 831 MetOne Instruments, Inc. Results and conclusion: The results from all the models clearly indicated that the winter season has the greatest influence on the bacterial growth, compared with the summer measured data. In the summer months there is no visible difference between the total air pollution outside the indoor air, but in the summer month’s microorganisms are more often present indicators of the presence of dust and fecal contamination. Air pollution with PM particles proportionally affects the microbiological contamination of indoor air. The reduction of air pollution is proportionally followed by a reduction of microbiological air contamination in both seasons and in both measured air samples. There is no visible association of microbiological contamination with the origin of increased air pollution, i.e. outside/indoor air.