Modeling the Impact of some Variables the COVID-19 Severe with CART Algorithm in Mashhad University of Medical Sciences

Considering that the new corona virus (COVID -19) is still prevalent, one of the important concerns isthe variables affecting the severity of the corona disease in the health of the society. In this study, the CART algorithmwas fitted to predict and determin the status of patients infected with the of COVID-19 in Mashhad University ofMedical Sciences.

This paper is a cross sectional-analytical study. Dataset were obtained from all of peoplereferred for the disease of COVID -19 collected at the Sinai system during the second peak and the fourth peak ofthe disease in Mashhad University of Medical Sciences. Data analysis was performed using JMP statistical softwareversion 13. Then for modeling, data mining methods and CART algorithm are used.

The descriptive findingsof our study showed that 6% of patients with positive PCR suffer from severe disease of COVID-19. The age variablewas very important in the severity of the disease. The age of 60 years old is the cut-off point for the severity ofthe disease, which increases the COVID-19 severe from about 3% under the age of 60 to about 18% over the age of60. The diseases of heart, kidney, respiratory, blood fat, and diabetes were other important variables.

The results of the CART model showed that for the age under 60 years the variables of heart disease, age, diabetes,respiratory disease, fat, gender and kidney, and for the age over 60 years the variables of age, heart disease, kidney,respiratory and diabetes were respectivly the most critical risk factors. According to the ROC curve, the fitted modelhas a good performance for COVID-19 severe disease, so that it increases up to 6 times the prediction of the COVID-19severe disease.