IMPLEMENTATION OF THE BAT ALGORITHM AND ENSEMBLE METHODS FOR TUBERCULOSIS DETECTION IN HIV PATIENTS BASED ON GENE EXPRESSION DATA Dr. Widyastuti Andriyani, S.Kom., M.Kom Additional Author HARNIANTI - 247110003 Primary Author mixed material bibliography Yogyakarta UNIVERSITAS TEKNOLOGI DIGITAL INDONESIA (UTDI) 2025 monographic Bahasa Inggris

Thesis

Tuberculosis (TB) remains one of the leading causes of mortality worldwide, particularly among individuals with Human Immunodeficiency Virus (HIV). Conventional diagnostic methods are often costly, time-consuming, and insufficiently sensitive for HIV patients. This study proposes a hybrid TB detection system based on gene expression data by integrating the Bat Algorithm for feature selection with ensemble learning classifiers (Random Forest, AdaBoost, and XGBoost). Using 44 peripheral blood RNA samples from the Gene Expression Omnibus (GSE50834), the Bat Algorithm effectively reduced dimensionality and improved classification performance. AdaBoost achieved the best results with an accuracy of 89% and an F1-score of 0.86. Although evaluation was conducted using an 80/20 train-test split, we acknowledge the need for more robust validation such as k-fold cross-validation or bootstrapping in future work. These findings highlight the potential of bioinformatics-driven approaches to enhance early TB detection in HIV patients and provide a promising complement to conventional diagnostics. Keywords— Tuberculosis, HIV, Gene Expression, Bat Algorithm, Ensemble learning Algoritma NONE Perpustakaan Universitas Teknologi Digital Indonesia Katalog Online Perpustakaan Universitas Teknologi Digital Indonesia 020 HAR I R.1 HAR0020MT0125 RAK REFERENSI TESIS MTI 020 HAR I R.1 18345 2026-02-20 08:22:27 2026-02-20 08:57:55 machine generated