APPLICATION OF C4.5 ALGORITHM WITH FEATURE SELECTION IN CLASSIFICATION OF DISCHARGE STATUS OF HEAD INJURY PATIENTS

Head trauma is a medical emergency that can cause brain damage and disability, leading to death. The discharge status of injured patients is classified into two: alive and dead. The purpose of this study is to apply the C4.5 algorithm without feature selection and by using Chi-Square and Mutual Information feature selection to show independent variables that significantly influence the discharge status of head injury patients. This research data is secondary data of patients who suffered head injuries at Dr. Abdul Aziz Hospital, Singkawang City, in 2019-2021. The independent variables used were age, gender, length of hospitalization, etiology of head injury, Suprasellar Cistern, and Glasscow Coma Scale, with the dependent variable being discharge status. Based on the study results, the Chi-Square feature selection results identified two variables that had a significant effect. In contrast, for the Mutual Information feature selection results, five variables had a significant impact on the dependent variable. The C4.5 Algorithm classification model without feature selection produces an accuracy of 88.57%, the C4.5 Algorithm classification model with Chi-Square feature selection produces an accuracy of 88.57%, and the C4.5 Algorithm classification model with Mutual Information feature selection produces an accuracy value of 91.42% with the highest accuracy obtained from the results of the C4.5 Algorithm model formation with Mutual Information feature selection.