Intellihealth: An Intelligent Medical Decision Support System Using A Novel Multi-Layer Classifier Ensemble Framework Based On Enhanced Bagging Approach With Multi-Objective Optimized Weighted Voting Scheme

Abstract

Decision support is a crucial function for decision makers in many industries. Typically, decision support systems (DSS) help decision-makers to gather and interpret information and build a foundation for decision-making. Medical Decision Support Systems (MDSS) play an increasingly important role in medical practice. By assisting doctors with making clinical decisions, DSS are expected to improve the quality of medical care. Conventional clinical decision support systems are based on individual classifiers or simple combination of these classifiers which tend to show moderate performance. In this thesis, we presented a novel multi-layer classifier ensemble framework based on enhanced bagging approach with multi-objective optimized weighted voting scheme for prediction of multiple diseases. The proposed model named ―HM-BagMoov‖ (Hierarchical Multi-level classifiers Bagging with Multi-objective Weighted voting) overcomes the limitations of conventional performance bottlenecks by utilizing an ensemble of seven heterogeneous classifiers: Naïve Bayes, Linear Regression, Linear Discriminant Analysis, K Nearest Neighbor, Support Vector Machine, Artificial Neural Network ensemble and Random Forest. The proposed ensemble framework utilizes different preprocessing techniques such as missing value imputation, feature selection, outlier detection and noise removal to improve the quality of data. Five different heart disease datasets, four breast cancer datasets, two diabetes datasets, two liver datasets and one hepatitis dataset are used for experimentation, evaluation and validation. The datasets are obtained from publicly available data repositories. Effectiveness of the proposed ensemble is investigated by comparison of results with several well-known classifiers as well as with ensembles. The experimental evaluation shows that the proposed framework dealt with all type of attributes and achieved high diagnosis accuracy. The f-ratio higher than f-critical and p-value less than 0.05 for 95% confidence interval indicate that the results are statistically significant for most of the datasets. Using HM-BagMoov we have developed an application called ―IntelliHealth‖ that may be used by hospitals/doctors for diagnostic advice.