Machine Learning-Based Real-time Intrusion Detection System for IoMT Networks

Abstract

In the evolving internet of things (IoT) era, ensuring robust security against adversarial attacks is of the utmost importance. The internet of medical things (IoMT) revolutionizes healthcare by enabling real-time patient monitoring and per-sonalized care through interconnected medical devices. However, this connectivity introduces significant security vulnerabilities, threatening data integrity, availability, and patient safety. To mitigate these risks, this research proposes a robust intrusion detection system (IDS) for IoMT that leverages advanced machine learning (ML) techniques. We develop a real-time intrusion detection model that effectively identifies and neutralizes diverse security threats. Our methodology involves extensive training and evaluation using the UNSW-NB15 dataset for comprehensive anomaly detection, encompassing both binary and multi-class classification. Furthermore, an initial exploration into medical data anomaly detection was conducted using a Kaggle Heart Disease dataset, providing foundational insights. By rigorously testing ML algorithms i.e., Random Forest, XGBoost, and Light-GBM, we optimize the IDS’s performance. By demonstrating confusion matrix, classification report, and accuracy, our model shows a significant improvement in the security and reliability of IoMT networks, safeguarding sensitive medical data and ensuring continuous, high-quality healthcare delivery. Our solution repre-sents a critical advancement in bolstering cybersecurity within the IoMT ecosystem.