A Malware Detection Approach For Internet of Medical Things Using Machine Learning

Abstract

The Internet of Medical Things (IoMT) revolutionizes healthcare by improving patient engagement, real-time monitoring, and data analysis through connected medical devices. However, the open and resource-constrained nature of IoMT networks introduces significant vulnerabilities, particularly to malware attacks that threaten the confidentiality, integrity, and availability of sensitive data and systems. These vulnerabilities pose serious risks to human life and the integrity of healthcare systems, making robust malware detection systems critical. Detecting unknown malware in IoMT environments remains a significant challenge for researchers and practitioners. Machine learning (ML) approaches have emerged as key tools for advancing intrusion detection, providing solutions to secure IoMT systems against evolving threats. This study proposes a low-cost, high-accuracy ML-based malware detection method for Windows PE files on medical devices and servers. Using the "Portable Executable (PE) malware" dataset, six ML algorithms were evaluated. Random Forest (RF) achieved the best results, with 99.28% accuracy and a processing time of 272 ms.