Abstract:

Abstract Insights into numerous biological processes and disease mechanisms are provided by microarray gene expression data, which is vital for biomedical research. Classifying samples into several predetermined groups based on their gene expression patterns is one of the core tasks in microarray data analysis. Our approach makes use of a thorough pipeline that includes feature selection, classification, and data preprocessing. To assure data quality and consistency, preprocessing procedures like normalization, missing value imputation, and noise reduction are first applied to the raw microarray data. The most insightful genes that considerably aid in the classification process are then found using a feature selection technique. We use a statistical class prediction approach based on an appropriate statistical model, such as logistic regression, support vector machines, or random forests, to carry out the classification. To ensure robustness and generalizability, the chosen model is trained on a labelled training set and its performance is assessed using cross-validation procedures. We carried out extensive tests on publically accessible microarray gene expression datasets related to various diseases to evaluate the efficacy of our suggested strategy. The outcomes show that our strategy outperforms previous approaches in terms of classification precision, sensitivity, specificity, and overall predictive power. Additionally, we discuss the biological significance of the discovered gene markers, offering light on putative molecular pathways causing the disorders under investigation.

Keywords: