Research Repository

An IoT-based smart healthcare system to detect dysphonia

Ali, Zulfiqar and Imran, Muhammad and Shoaib, Muhammad (2021) 'An IoT-based smart healthcare system to detect dysphonia.' Neural Computing and Applications. ISSN 0941-0643

[img] Text
Samart_Healthcare_NCA_final_accepted.pdf - Accepted Version
Restricted to Repository staff only until 2 January 2022.

Download (1MB) | Request a copy

Abstract

Smart healthcare systems for the internet of things (IoT) platform are cost-efficient and facilitate continuous remote monitoring of patients to avoid unnecessary hospital visits and long waiting times to see practitioners. Presenting a smart healthcare system for the detection of dysphonia can reduce the suffering and pain of patients by providing an initial evaluation of voice. This preliminary feedback of voice could minimize the burden on ENT specialists by referring only genuine cases to them as well as giving an early alarm of potential voice complications to patients. Any possible delay in the treatment and/or inaccurate diagnosis using the subjective nature of tools may lead to severe circumstances for an individual because some types of dysphonia are life-threatening. Therefore, an accurate and reliable smart healthcare system for IoT platform to detect dysphonia is proposed and implemented in this study. Higher-order directional derivatives are used to analyze the time–frequency spectrum of signals in the proposed system. The computed derivatives provide essential and vital information by analyzing the spectrum along different directions to capture the changes that appeared due to malfunctioning the vocal folds. The proposed system provides 99.1% accuracy, while the sensitivity and specificity are 99.4 and 98.1%, respectively. The experimental results showed that the proposed system could provide better classification accuracy than the traditional non-directional first-order derivatives. Hence, the system can be used as a reliable tool for detecting dysphonia and implemented in edge devices to avoid latency issues and protect privacy, unlike cloud processing.

Item Type: Article
Uncontrolled Keywords: Edge analytics, Higher-order local derivatives, Vocal folds disorders, Spectrum analysis, Support vector machine (SVM)
Divisions: Faculty of Science and Health > Computer Science and Electronic Engineering, School of
Depositing User: Elements
Date Deposited: 25 Jan 2021 15:24
Last Modified: 25 Jan 2021 15:24
URI: http://repository.essex.ac.uk/id/eprint/29617

Actions (login required)

View Item View Item