Abstract

*Yashwant Kumar Kolli and Karthick M.

ABSTRACT

Internet of Medical Things (IoMT) has been employed in healthcare in combination with deep learning and cloud computing, which has proved to be useful for real-time monitoring and accurate prediction of diseases. On the other hand, traditional healthcare systems have several challenges like delay in response to emergency situations, high false alarm rates, and privacy issues. A secure and effective COVID-19 detection system using ResNet-101 in combination with Sharpness-Aware Minimization (SAM) optimization has been proposed in this research. This system is used for detecting COVID-19 by processing the chest X-ray image for improving the diagnostic accuracy while considering the problems of overfitting and generalization. Also, federated learning will be used to ensure data privacy. The AWS cloud infrastructure is used for the purpose of managing resources and scaling. A system with high performance has been developed by obtaining 98.23% accuracy, 97.89% precision, 98.15% recall, and 97.90% F1-score. Inference time analysis has been conducted for analysing the time efficiency of the system for different batch sizes. The above system reduces the number of false positives and improves the diagnosis at the initial stage and is robust. In the future, additional work will be carried out to include other datasets, multi-modal medical imaging, and real-time deployment for larger clinical applications.