A novel optimised apnea classification with AA-CNN method by utilising the EDR and ECG features. (13th July 2022)

• Record Type: Journal Article
• Authors: Smruthy, A.; Suchetha, M.
• Is Part Of: International journal of ad hoc and ubiquitous computing. Volume 41:Number 1(2022)
• Page Start: 58
• Page End: 67
• Publication Date: 2022-07-13
• Subjects: two-level variational mode decomposition -- ECG derived respiratory signal -- electrocardiogram -- ECG -- support vector machine -- convolution neural network -- CNN -- multi-features; Ubiquitous computing -- Periodicals; Embedded computer systems -- Periodicals; Electronic data processing -- Distributed processing -- Periodicals; Wireless communication systems -- Periodicals; Computer architecture -- Periodicals; 004.2
• Languages: English
• ISSNs: 1743-8225

Abstract

Convolution neural network (CNN) has shown promising growth in recent years. The main reason for the above growth is the highest classification accuracy achieved within a short span of time. However, the traditional CNN architecture limits the fixed window size of the convolution filter. Therefore the architecture fails to learn multiple features properly. In this scenario, we propose an adaptive attention convolution neural network (AA-CNN), which is able to learn multi-features. The proposed work is divided into two stages. In the first stage, electrocardiogram (ECG) and ECG-derived respiratory signal (EDR) were extracted using a novel two-level variational mode decomposition algorithm. In the next stage, the optimal convoluted features were derived using the AA-CNN architecture. To validate our proposed work, we have developed an apnea classification system by using the concept of AA-CNN and the support vector model. The overall accuracy of 98.18% is obtained for our proposed work.