An IoT enabled secured clinical health care framework for diagnosis of heart diseases. (February 2023)

• Record Type: Journal Article
• Authors: Raheja, Nisha; Kumar Manocha, Amit
• Is Part Of: Biomedical signal processing and control. Volume 80:Part 2(2023)
• Publication Date: 2023-02
• Subjects: Arrhythmias -- Triple data encryption standard (3-DES) -- Water cycle optimization (WCO) -- IoT -- Deep learned convolutional neural network -- ECG -- Classification; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.104368
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: This research work proposed a framework for IoT enabled ECG monitoring and cardiac arrhythmia classification model. Proposed method could save lives of rural area people particularly in Golden Hour. Pre-processing of ECG signal is performed using SG filter and MOWPT followed by delineation process. For encryption, the 3-DES is used and the WCO has been used for authentication. The proposed deep learned CNN classifier achieved accuracy of 99.12%, 100% sensitivity and 99.9% specificity for MIT-BIH dataset. Abstract: World Health Organization (WHO) estimated cardiovascular diseases (CVDs) is main contributor to the death count globally. In addition, there are shortcomings of specialists in developing countries to handle these diseases. Internet of things (IoT) and significant advances in machine learning techniques offers more efficient healthcare services. The research work proposed a IoT enabled, cloud-centric solution for remote monitoring of ECG, which could save lives of rural area people particularly in Golden hour. Firstly, pre-processing of ECG signal is performed using Savitzky-Golay (SG filter) for removal of the baseline wanders and maximum overlap discrete wavelet packet transform (MOWPT) is applied for the removal of high noise followed by delineation process of ECG characteristic points. Then deep learned convolution neural network (CNN) is trained and classification is done. For encryption and authentication, Triple data encryption standard is used for security whereas, water cycle optimization technique is used for authentication. IoT is implemented using the ThingSpeak platform, which enables the visualization and analysed the ECG data over the cloud server. Using ThingSpeak, encrypted and authenticated ECG data is shared with cardiologist for examination. To validate the proposed classification method, the standard annotated dataset of the MIT-BIH database is tested and various performance metrics were calculated. The proposed deep learned CNN classifier classifies the heartbeats into five types of arrhythmias with an average accuracy of 99.12%, whereas the sensitivity and specificity of this proposed model are 100% and 99.9% respectively. The achieved results demonstrated better performance as compared to methods reported in the literature.