Clinical decision support system for neuro-degenerative disorders: An optimal feature selective classifier and identification of predictor markers

Abstract

Detecting divergence between Neuro-degenerative diseases is essential for right treatment. This intelligent system is implemented through computational methods to predict the class of Neuro-degenerative disease (Alzheimer’s, Parkinson’s or common) from the structural and physicochemical properties (1437 attributes respectively) of protein sequences extracted from genes. The Gene Set Enrichment Analysis database (GSEA db) was utilized to obtain the gene sets that contributed to the development of Alzheimer’s and Parkinson’s disease. Optimal features for classification were obtained by applying Gain Ratio followed by Correlation-based Feature Selection (CFS) and Decremental Feature Selection (DFS) on extracted properties from Kyoto Encyclopedia of Genes and Genomes (KEGG) dataset for the GSEA database. The selected features are evaluated using Random Forest model. The Clinical Decision Support System (CDSS) was build which extract rules from the least sized Decision tree automatically and predict the type of Neuro-degenerative disorder as Alzheimer’s disease, Parkinson’s disease or common to both diseases. The CDSS predicts the disease with classification accuracy as 79.7% and Mathew’s Correlation Coefficient as 0.689.