Attack Detection in Internet of Medical Things Through Ensemble Machine Learning Models

Abstract

The Internet of Medical Things (IoMT) revolutionizes healthcare by enabling real‐time patient monitoring and treatment. Still, its vulnerability to cyberattacks poses significant risks to patient safety and privacy. This study leverages machine learning (ML) to enhance IoMT security, evaluating conventional and ensemble models on the CIC IoT and the IoMT datasets. Models are trained using all features, and a reduced feature set is selected via feature importance. Ensemble models, particularly the Stacked Ensemble (SE), outperform traditional ML, achieving detection rates of 99% on CIC IoT and 97.58% on IoMT, with lower false positives and reduced execution times. By improving accuracy and resource efficiency, this approach strengthens IoMT security, safeguarding patient data and enhancing healthcare outcomes.