A model to find optimal percentage of training and testing data for efficient ECG analysis using neural network

Electrocardiogram (ECG) data is one of the most important physiological parameter for detecting heartbeat, emotions and stress levels of patients. The problem is to develop a model that can diagnose an ECG data efficiently with higher accuracy overtime. In this paper, Authors have proposed a model that identifies the percentage division of data so as to get the maximum possible accuracy for a particular dataset. For experimental purpose, the authors have used neural networks for the analysis of the standard and raw data taken from MIT-BIH long-term ECG database using R as a platform. The database is divided into different ratios of training and testing data and the model is trained to attain the best percentage division of the particular patient’s data based upon its accuracy.