A Classification of Inflammatory Bowel Disease using Ensemble Learning Model

Inflammatory Bowel Disease (IBD) is an enduring, intermittent bowel disorder with unidentified mechanism and etiology. It typically contains Ulcerative colitis (UC) and Crohn’s disease (CD) that are categorized by several bowel ulcers. An impedance phase displaces dynamic development, accumulating susceptibility to exterior pathogens and theoretically leading to neurotic symptoms which encompasses bowel blockade, intestinal tear and liver lesions, with high disability ratio. The existing models have more overfitting, and class imbalance problems. To overcome that, an Ensemble Learning (EL) model such as Logistic Regression, Random Forest and Gaussian Naive Bayes is proposed in this research. Initially, the collected Gene Expression Series 106817 (GSE106817) dataset is preprocessed with Recurrent MultiArray Average (RMA) Normalization and Log2 Data Transformation to eliminate the noises. Next, Recursive Feature Elimination (RFE) technique is used to reduce high-dimensional data for selecting optimal features; and finally, the proposed EL model gives a precise rate in detecting and classifying IBD as CD, UC and Healthy controls. The proposed EL model gives better results than the other compared existing methods Random Forest-Support Vector Machine (RF-SVM) in the context of recall, accuracy, f1 score, and precision as 99.38%, 99.75%, 94.72%, and 97.94% respectively.