Using Machine Learning to Improve Detection of Cyberattacks Against the Internet of Medical Things (IoMT)
Open Access DepositedThe Internet of Medical Things is a subset of the Internet of Things vulnerable to cyberattacks. This poses risks to patient safety and creates legal challenges for healthcare organizations. Strengthening this essential infrastructure is crucial to maximizing the benefits of IoMT technology while minimizing risks. Protecting IoMT involves identifying potential threats. This study introduces novel machine learning methods for threat categorization, utilizing Convolutional Neural Network, Decision Tree, ExtraTrees Classifier, Multi-Layer Perceptron, Random Forest, and XGBoost model architectures. This research provides novel contributions to artificial intelligence by enhancing model explainability and adding insight for dimensionality reduction through Shapley Additive Explanations. It also improves the identification of low-support classes over existing models by implementing Stratified K-Fold cross-validation to account for class imbalance. Through exploratory data analysis, two key issues were identified. The first was feature imbalance, resolved using Stratified K-Fold cross-validation. The next issue was high data dimensionality, which led to unexplainable model outcomes. This was tackled by incorporating Shapley Additive Explanations to validate and incorporate Explainable AI. An extra benefit of Shapley Additive Explanations is informing dimensionality reduction for high-dimensional datasets. These models and feature engineering processes present a distinctive method for detecting IoMT attacks, with the Stratified K-Fold Random Forest Classifier emerging as the most effective model, exhibiting minimal overfitting issues. This research directly contributes to intrusion detection and prevention systems, enabling the detection and prevention of threats. These models hold the potential for adaptation to online and near-real-time applications, representing an advancement in cyberattack detection and disruption. Overall, this research enhances our comprehension of the risks associated with IoMT and provides insights into how medical facilities can effectively mitigate these threats.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.