Coronavirus Infectious Disease (COVID-19) Modeling: Evidence of Geographical Signals

Abstract

2019 Novel Coronavirus (2019-nCoV) is a virus (more specifically, coronavirus) known as the cause of respiratory disease outbreak that was first detected in Wuhan, China. In the present study, we demonstrated the applicability of four different data mining techniques namely Decision Tree (DT), Random Forests (RF), Logistic Model Trees (LMT) and Naive Bayes (NB) classifiers to model and present the development of Coronavirus disease (COVID-19) based on 482 records of cases. Johns Hopkins University has created an outstanding database using the data from the affected cases (Johns Hopkins Github repository). Data were obtained from the google sheets associated and are available from 22 Jan, 2020. Owing to input factors such as latitude, longitude, age, sex and living in Wuhan, the results confirmed that the accuracy of the models is typically between 0.8744 and 0.9083 using the kappa statistics. Overall, these findings may lead to more information about the role of environmental factors and the spread of COVID-19 disease seasonal dynamics with respect to the effects of latitude and longitude. Evidence of geographical signals has been found and further works on COVID-19 based on geographical signals and seasonal dynamics are strongly recommended.

Note: Funding: None.

Conflict of Interest: None.

Keywords: COVID-19, Data Mining, Classifiers, Latitude, Longitude