An Intelligent Technique for Predicting Adverse Drug Reactions (ADRs) Using Modified Shrinkage Function Based Extreme Gradient Boost (MSXGB) Classifier

1C. Rathika and Dr.V. Umarani


It has turned out to be highly significant for the prediction of Adverse Drug Reactions (ADRs) in consequence of the massive universal health troubles and malfunction of drugs. This signifies the necessity for earlier prediction of possible ADRs in preclinical phases that could recover failures of drug and decrease the duration and expense of progress and therefore offering effective and harmless therapeutic decisions for patients. Despite the fact that different techniques have been proposed for prediction of ADR, still there exists need to enhance the classification and precision in predicting adverse drug reactions. This research work presents a framework for discovering the side effects using the best feature selection and classification methods. The proposed research work presents a new approach to measure the features (attributes) in drug prediction dataset using the Modified Orthogonal Opposition Learning based Cuckoo Search Algorithm (MOOLCSA). Finally in this work, Modified Extreme Gradient Boost (MXGB)is used as a classifier that acts as a supervised machine learning approach and could predict Adverse Drug Reactions. This proposed approach holds the benefits of better predictability, interpretability and is an intelligent technique that is beneficial for both patients and also the medical researchers. The simulation result shows that the proposed Modified Shrinkage function based Extreme Gradient Boost (MSXGB) classifier model provides best side effect prediction compared to the other prediction techniques.


Drug Side Effect, Predictive Modeling, Machine Learning, Feature Selection, Modified Orthogonal Opposition Learning based Cuckoo Search Algorithm (MOOLCSA), Classification, Modified Shrinkage Function based Extreme Gradient Boost (MSXGB) Classifier Model.

Paper Details
IssueIssue 4