An Improved Data Classification in Edge Cloud-Assisted IoMT: Leveraging Machine Learning and Feature Selection

As data generated by the IoMT devices are offloaded to the cloud, processing and analyzing such large-scale data to extract useful information presents challenges in terms of latency, bandwidth, and privacy. Edge computing has emerged as a promising paradigm to address these challenges. For this purpose, the proposed approach explores the benefits of performing Feature Selection methods, including filters, wrappers, and embedded techniques, along with ML algorithms at the edge. As a result, a novel framework is proposed that requires an optimal number of features from the dataset to build an optimal data classification model. The results of the study show that feature selection can significantly improve the classification accuracy and performance of ML algorithms when applied to the medical dataset. Moreover, the simulation results confirm that the XGBoost classifier utilizing the ET algorithm achieved the highest classification accuracy of 95% surpassing the current state-of-the-art.