Classification and Prediction of severity of Inflammatory Bowel Disease using Machine Learning









Abstract

This paper shows a novel approach to classifying the status of Inflammatory Bowel Disease from vitamin D in children and adolescents. IBD is a phrase used to refer to gastrointestinal system swelling that is recurrent. It is observed that low vitamin D levels are linked with higher risk, particularly colon cancer in people with IBD. Vitamin D can play a protective role on gut health if started earlier in patients with IBD. Machine learning has many advantages in the healthcare industry as it can predict the presence of a condition much before and efficiently and is found helpful for doctors to suggest better treatment to a patient with the assistance of artificial intelligence. This paper uses an open dataset for the analysis of patients with IBD and their corresponding vitamin D level of Serum 25(OH) D concentration. The data is classified based on severity index into three classes as low risk, moderate, and high-risk patients. Tree classifiers, Support Vector Machine (SVM), and ensemble boosted tree classifiers are used for training and comparative analysis is done. Dataset consists of 31 features which include healthy and IBD patients in the age range of 2 to 20 years. The classification accuracy is maximum for ensemble trees classifier 98% and Area under ROC curve is 0.98.


Modules


Algorithms


Software And Hardware