Indonesian Journal of Electrical Engineering and Computer Science Vol. 21, No. 3, March 2021, pp. 1530~1539 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v21.i3.pp1530-1539  1530 Journal homepage: http://ijeecs.iaescore.com Predicting heart failure using a wrapper-based feature selection Minh Tuan Le 1 , Minh Thanh Vo 2 , Nhat Tan Pham 3 , Son V.T Dao 4 1,2 SEE, International University, Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Viet Nam 3,4 SIEM, VNU-International University, Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Viet Nam Article Info ABSTRACT Article history: Received Oct 30, 2020 Revised Dec 7, 2020 Accepted Dec 23, 2020 In the current health system, it is very difficult for medical practitioners/ physicians to diagnose the effectiveness of heart contraction. In this research, we proposed a machine learning model to predict heart contraction using an artificial neural network (ANN). We also proposed a novel wrapper-based feature selection utilizing a grey wolf optimization (GWO) to reduce the number of required input attributes. In this work, we compared the results achieved using our method and several conventional machine learning algorithms approaches such as support vector machine, decision tree, K- nearest neighbor, naïve bayes, random forest, and logistic regression. Computational results show not only that much fewer features are needed, but also higher prediction accuracy can be achieved around 87%. This work has the potential to be applicable to clinical practice and become a supporting tool for doctors/physicians. Keywords: Feature selection Grey wolf optimization Heart failure Multilayer perceptron Neural network This is an open access article under the CC BY-SA license. Corresponding Author: Son V.T Dao School of Industrial and Management International University Vietnam National University HCMC Ho Chi Minh City, Viet Nam Email: dvtson@hcmiu.edu.vn 1. INTRODUCTION The term "heart failure" referring to a condition in which the heart's contraction is not as effective as it should be. The heart is a vital organ in the human body because it pumps blood to every other organ. A patient who is living vegetative states still needs the heart to survive. Heart failure (HF) is a chronic condition in which one of the ventricles or atriums on both sides is not able to pump rich oxygen into the body and poor oxygen into the lungs. There are several common reasons cause heart failure. The majority of (HF) patients are elderly. Cardiac arrest often gradually and deliberately develops after parts of a heart got weaken and makes others such as ventricles and atriums do extra workloads to provide enough blood and oxygen to the body [1-2]. With the ubiquitous application of technology in the medical field, it helps the cost of diagnosis to be inexpensive. Unfortunately, nowadays the number of patients who have been diagnoses with heart failure is gradually increasing in suburbs and dramatically increasing in urban areas. Therefore, the earlier for getting diagnosed, the better off it will be for the patients. Because of the difficulty of diagnosing the process of a heart failure condition, it might cause a postponement in treatment operation. Therefore, it is crucial to develop a heart disease prediction system for heart failure to support whoever works in the medical professional field to diagnose patients with conditions more rapid and accurate. Deep learning and Machine learning algorithms have been successfully applied to various field [3-4], especially medical field to support doctor/physician to diagnose various diseases such as heart failure, diabetes. ANN has also been applied by researchers in the medical field [5-7].