ABSTRACT
Presently, wearables act as a vital part of healthcare sector and they are able to offer exclusive perceptions about the person's health conditions. In contrast to traditional diagnosis in a hospital environment, wearables can give unrestricted access to real-time physiological data. COVID-19 epidemic is increasing at a faster rate with limited test kits. Hence, it becomes essential to develop a novel COVID-19 diagnostic model. Numerous studies were based on the utilization of artificial intelligence techniques on radiological images to precisely identify the disease. This chapter presents an efficient fusion-based feature extraction with multikernel extreme learning machine (FFE-MKELM) for COVID-19 diagnosis using internet of things (IoT) and wearables. Primarily, the wearables and IoT are used to capture the radiological images of the patient. The presented FFE-MKELM model incorporates Gaussian filtering based preprocessing for removing the noise that exists in the radiological image. Besides, directional local extreme patterns with deep features based on Inception v4 model are applied for the FFE process. In addition, MKELM model is utilized as a classification model to determine the appropriate class label of the input radiological images. Moreover, monarch butterfly optimization algorithm is applied to fine tune the parameters involved in the MKELM model. Experimental validation of the FFE-MKELM model is performed against benchmark dataset and the outcomes are inspected under different measures. The resultant simulation outcome ensured the betterment of the FFE-MKELM method by demonstrating an increased sensitivity of 97.34%, specificity of 97.26%, accuracy of 97.14%, and F-measure of 97.01%. © 2022 Elsevier Inc. All rights reserved.