Deep learning based classification for healthcare data analysis system

Abstract

This paper presents a deep learning based mechanism to analyze the healthcare data to detect the possible anomalies and classify the data into different so that we can know the nature of health problem. An implementation of deep convolutional neural network (DCNN) to classify the image patterns data extracted from electrocardiograph (ECG) is discussed in detail. A dedicated convolutional neural network will be trained using different data samples taken from various patients termed as training data. On later stage, the algorithm is tested using test data samples and it is observed that the proposed algorithm does perform efficient, stable and superior classification performance for the detection of normal beats (N-Type), ventricular ectopic beats (V-Type) and super ventricular ectopic beats (SV-Type). The experimental analysis shows the recognition accuracy and loss value. Subsequently, sensitivity and specificity of the algorithm is measured to show the effectiveness of the proposed solution.