This paper presents a novel hybrid intelligent system based on ensemble of neuro-fuzzy classifiers (NFCs) and rough set theory for automatic detection of heart disease. A pool of NFCs which is trained using scale conjugate gradient is generated. Rough sets are used to identify the most significant classifiers which contributed in the committee. Consequently, the classifiers space of the ensemble is reduced which in turn reduce the complexity of the problem. The results of the reduced NFCs are combined by majority voting rule to obtain the final diagnosis decision. The proposed system is applied on the dataset taken from the well- known Cleveland heart disease database. Performance measures such as accuracy, specificity, sensitivity, F-score which are commonly used in medical diagnosis were evaluated to convey the qualities of the proposed method. The results obtained showing the efficiency of the proposed hybrid combined system.
Digital Object Identifier (DOI)
H. El-Baz, A.
"Neuro-Fuzzy Ensemble Model-Based Rough Set Classifier Selection for Automatic Detection of Heart Disease,"
Applied Mathematics & Information Sciences: Vol. 12:
2, Article 5.
Available at: https://digitalcommons.aaru.edu.jo/amis/vol12/iss2/5