Improved Interpretation Model for Heart Disease Diagnosis Using Artificial Neural Networks
Dr.F. Syed Anwar HussainyDepartment of Computer Science and Engineering (AI & ML), Faculty of Engineering and Technology, Jain (Deemed to be University), Bangalore syedanwar.f@jainuniversity.ac.in0000-0003-2837-3856
J. JayapradhaAssistant Professor, SRM Institute of Science and Technology, SRM Nagar, Kattankulathur jayapraj@srmist.edu.in0000-0002-2548-9135
Dr. Mardeni RosleeProfessor, Faculty of Engineering, Multimedia University, Cyberjaya mardeni.roslee@mmu.edu.my0000-0001-8250-4031
Dr.T. Senthil KumarAssociate Professor, SRM Institute of Science and Technology, SRM Nagar senthilt2@srmist.edu.in0000-0002-2200-3339
Dr. Chilakala SudhamaniPostdoctoral Research Fellow, Faculty of Engineering, Multimedia University, Cyberjaya sudhamanich@gmail.com0000-0002-8823-9053
Azmi IsmailCentre of Excellence for Intelligent Network, Telekom Malaysia Research & Development Cyberjaya, Malaysia. azmi@tmrnd.com.my0000-0001-5792-9601
Anwar Faizd OsmanSpectre Solution Sdn Bhd anwarfaizd@spectresolution.com0000-0001-9563-9138
Dr. Fatimah Zaharah AliUniversiti Teknologi MARA, Faculty of Electrical Engineering, Shah Alam fatimah_zaharah@uitm.edu.my0000-0002-5467-4049
Idris Olalekan AdeoyeCentre for Wireless Technology, Faculty of AI & Engineering, Multimedia University Cyberjaya, Malaysia. idreezleks@gmail.com0009-0008-3450-7519
In the present scenario, health care is a predestined process to be considered in human life. While heart diseases are concerned, cardiovascular disease (CVD) is a wide class of diseases that damages blood vessels and the heart. In the medical field, huge health data are available to study and process; hence, machine learning methods are required for appropriate decision-making, specifically in terms of heart disease prediction and diagnosis. For enhancing the appropriation rate of decision-making in CVD diagnosis, this paper proposes an Improved Interpretation Model for Heart Disease Diagnosis (IIM-HDD) using Artificial neural networks. The model incorporates data acquisition, pre-processing, feature selection, training, and testing for diagnosis. For training and validation, the data from benchmark datasets are combined and used. Moreover, feature selection is computed with a relief-based selection process. The ANN model is trained to produce the output, corresponding to their input features. The results computations are processed with metrics that include classification, accuracy, precision rate, error rate, specificity, sensitivity, F1 score, and other comparisons are also provided for proving the proposed model. The results show that the work outperforms the other compared models in respective metrics.