Machine learning based intrusion detection system for IoMT

Millennials have the advantage of accessing readily available modern scientific advancements, particularly in technology. One of these technologies that encompasses varied functionalities is the Internet of Things (IoT). In the midst of the Covid-19 pandemic, IoT, specifically Internet of Medical Things (IoMT), had pivotal significance in monitoring and tracking different health parameters. It autonomously manages an individual's health data and stores the same as Electronic Health Records (EHRs). However, the networking protocols used by IoMT are not adequate enough to ensure the security and privacy of EHRs. Consequently, such technology is susceptible to cyber-attacks, which have become more prevalent over time and have taken various forms, that generally the stakeholders are not aware of. This paper introduces machine learning-driven intrusion detection systems as a solution to tackle this issue. The focus of this study is on devising a Machine Learning (ML) oriented Intrusion Detection System (IDS) designed to identify cyber-attacks targeting IoMT based systems. Several classification based ML techniques such as Multinomial Naive Bayes, Logistic Regression, Logistic Regression with Stochastic Gradient Descent, Linear Support Vector Classification, Decision Tree, Ensemble Voting Classifier, Bagging, Random Forest, Adaptive Boosting, Gradient Boosting and Extreme Gradient Boosting were used, whereupon the Adaptive Boosting was experimentally found to perform the best on performance metrics such as accuracy, precision, recall, F1-score, False Detection Rate (FDR) and False Positive Rate (FPR). Further, it was found that Adaptive boosting based IDS for IoMT performed comparatively better than the existing ToN_IoT based IDS models on performance metrics such as accuracy, F1-score, FPR and FDR.