Machine Learning Framework for the Prediction of Alzheimer’s Disease Using Gene Expression Data Based on Efficient Gene Selection

In recent years, much research has focused on using machine learning (ML) for disease prediction based gene expression (GE) data. However, many diseases have received considerable attention, whereas some, including Alzheimer’s (AD), not, perhaps due to data shortage. The present work is intended fill this gap by introducing a symmetric framework predict AD from GE data, with the aim produce most accurate smallest number of genes. works in four stages after it receives training dataset: pre-processing, selection (GS), classification, and prediction. symmetry model manifested all its stages. pre-processing stage columns dataset are pre-processed identically. GS stage, same user-defined filter metrics invoked every individually, so wrapper metrics. classification ML models applied identically minimal set genes selected preceding stage. core proposed meticulous algorithm which we designed nominate eight subsets original provided dataset. Exploring subsets, selects best one describe AD, also subset. For credible results, calculates performance repeated stratified k-fold cross validation. To evaluate framework, used an 1157 cases 39,280 genes, obtained combining smaller public datasets. were split two partitions, 1000 training/testing, 10-fold CV 30 times, 157 From testing/training phase, identified only 1058 be relevant support vector (SVM) these final validation, that never seen SVM classifier. evaluation, evaluated classifier via six metrics, impressive values. Specifically, 0.97, 0.98, 0.945, 0.972, 0.975 sensitivity (recall), specificity, precision, kappa index, AUC, accuracy, respectively.