Evergreen I
Random Forest Classification and Biomarker Discovering Model to Detect HIV-Associated Neurocognitive Impairment (306307)
Bryan Martinez, University of Texas Rio Grande Valley*Hansapani S Rodrigo, University of Texas Rio Grande Valley
Upal Roy, University of Texas Rio Grande Valley
Keywords: HIV, Neurocognitive impairment, Random Forest, Biomarker Identification
HIV- associated neurocognitive disorders (HAND) occur frequently among people with end-stage Acquired Immunodeficiency Syndrome (AIDS). Early intervention and detection enhance the possibility of reducing the risk of suffering from severe consequences due to HAND. Genome-wide screening of transcription regulation in brain tissue help in identifying substantial abnormalities present in patients' gene transcripts and to discover possible biomarkers. This study explores the possibility of developing a classification model using a random forest to identify the risk of diagnosing with different forms of HAND (HIV associated Dementia and HIV-encephalitis). Random forest is a popular tree-based ensemble machine learning tool that is highly data adaptive which account for correlation as well as interactions among features. Gene expression levels of three different brain sectors; white matter, neocortex and neostriatum from each patient have been utilized in developing our model. The developed model has a 65% accuracy, 69% precision and 67% recall rates. The heatmap reveals significant expression level differences in certain genes of the patients relative to controls, unfolding favorable