Performance Analysis of SMOTE and Feature Selection Methods in Traumatic Brain Injury Stage Classification
Year: 2026
Type: conference-paper
Abstract: Traumatic brain injury (TBI) is a medical condition requiring urgent intervention to preserve vital functions and prevent permanent damage. Machine learning approaches make it possible to classify the severity of TBI more quickly and objectively than with current methods. In this study, two datasets comprising 3 and 4 classes were created using data from 1,228 patients collected by the Department of Neurosurgery at Trakya University. The SMOTE method was used to address class imbalance in the dataset, and feature selection algorithms were employed to identify subsets of features that were more discriminatory. For model development, k-Nearest Neighbor (kNN), Random Forest (RF), Support Vector Machine (SVM), and Multilayer Perceptron (MLP) models were run and their performances were compared. It was observed that the SVM model achieved the highest accuracy success rate of %90.65 with feature selection.
Institution: Trakya University
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7202225813