Artificial Intelligence Application for the Classification of Central Nervous System Tumors Based on Blood Biomarkers

This study addresses the imperative task of Central Nervous System (CNS) tumor classification through the amalgamation of artificial intelligence (AI) and data mining methodologies. Leveraging microarray data, our approach involves feature selection algorithms to navigate the intricacies of gene expression patterns, enabling the identification of markers pivotal in neuro-oncology. The dataset, encompassing 60 samples with 7129 features, undergoes a rigorous analysis employing diverse data mining techniques and machine learning algorithms, including K-Nearest Neighbors, Support Vector Machines, Linear Discriminant Analysis, Decision Tree for Classification, and Naive Bayes. We introduce the Hybrid Selection Method, a three-phase algorithm merging filter and wrapper techniques, aiming for optimal classification accuracy while minimizing attributes. Noteworthy findings showcase enhanced accuracy and reduced gene subsets, particularly in KNN and SVM models. These outcomes underscore the Hybrid Selection Method's potential in advancing gene selection for CNS tumor classification, promising improvements in neuro-oncology diagnostics and personalized treatment strategies.