An Intrusion Detection System For Internet of Medical Things Using Machine Learning Approaches

Abstract

The Internet of Medical Things (IoMT) is transforming healthcare by increasing accuracy, reliability, and scalability through smart, connected medical devices that enable real-time monitoring and accurate diagnosis. However, due to their resource limitations and diverse features, IoMT networks face significant privacy and security challenges, including vulnerabilities to cyber threats such as man-in-the-middle, DoS, remote hijacking, and replay. The healthcare ecosystem faces critical risks from IoMT vulnerabilities, threatening human lives and system integrity. To address evolving cyber-attacks, intrusion detection systems (IDS) have become essential, with machine learning (ML) techniques proving highly effective. This study presents a low-cost, high-accuracy ML-based attack detection framework for securing IoMT devices. Eight ML models, including Decision Tree, Support Vector Machine, Naive Bayes, Gradient Boosting, K-Nearest Neighbor, Random Forest, and XGBoost, were evaluated on the IoT-Healthcare security dataset. XGBoost emerged as the top performer, achieving 99.98% accuracy in just 233 ms.