Machine learning for Markov modeling of COVID-19 dynamics concerning air quality index, PM_-2.5, NO₂, PM_-10, and O₃

Abstract

In this research machine learning technique is adopted for Markov modelling of the COVID-19 dynamics with variations in Air Quality Index AQI, PM_-2.5, NO₂, PM_-10, and O₃, respectively. Thus the long-run disease dynamics of the COVID-19 are studied concerning the Air Quality Index (AQI), PM_-2.5, NO₂, PM_-10, and O₃, respectively, by using eigen space decomposition. Data of the Chhattisgarh state of India is taken into consideration and the study analyzed in two phases. In Phase-1 the time duration is from March 15, 2020, to May 01, 2020, and for Phase-2 the time duration was from Jun 01, 2020, to Jul 15, 2020. Change in COVID-19 related to AQI showed that initially when the AQI changed from 103 to 84.83 the disease dynamics also changed, and the first cases of COVID-19 were reported. In the next two fortnights from March 15, 2020, and April 01, 2020, the dynamics were the same, latter the AQI changed from 84.83 to 63.83, but this change does not affect the disease dynamics in long run from April 15, 2020, to Jul 15, 2020. In phase 1 the solution obtained shows a cyclic trend with initially decreasing, then increasing, and again a decreasing trend for changes concerning PM_-2.5. The disease dynamics concerning PM_-2.5, NO₂, PM_-10, and O₃, respectively, based on initial transition showed the same trend for PM_-2.5, NO₂, and PM_-10. Moreover, for O₃ the disease dynamics were found different than the other three parameters. COVID-19 showed a negative correlation with AQI, PM-2.5, NO2, PM-10. Moreover a positive correlation with O₃. This proved that the lockdown and ban on transport activities improved AQI, PM-2.5, NO2, PM-10, but not O₃. The research exploits the learning capability of renowned Python machine learning module sklearn to solve the Markov model. The findings prove that Markov modeling is promising method for planning control measures for disease with compatibility to air quality alterations.

Rights

http://creativecommons.org/licenses/by/4.0/

Cite as

Downloadable citations

Identifiers

DOI

https://doi.org/10.18280/ijcmem.120202

Repository URI

https://research-portal.uws.ac.uk/en/publications/214f234b-eb6e-4852-b896-c63d409dcff5