FOID: A Feature-Optimized Intrusion Detection System for Securing IoMT Healthcare Networks
Jamshed Ali Shaikh, Chengliang Wang, Muhammad Owais, Uzair Zia, Muhammad Wajeeh Us Sima, Muhammad Arshad
Abstract
The rapid integration of interconnected healthcare devices, software, operating systems, and networks within the Internet of Medical Things (IoMT) has greatly advanced healthcare delivery. However, this interconnectivity also presents significant security challenges. Many IoMT devices lack robust security mechanisms, making them vulnerable to intrusion and cyber-attacks. To address these vulnerabilities, we propose FOID, a feature-optimized intrusion detection scheme (IDS) designed specifically for securing IoMT healthcare systems. FOID enhances the security and reliability of healthcare networks by efficiently detecting and mitigating security breaches, particularly unauthorized access attempts. The scheme utilizes recursive feature elimination (RFE) alongside advanced machine learning models and integrates the SHapley Additive exPlanations (SHAP) mechanism to provide transparency. This comprehensive approach ensures both the integrity and confidentiality of healthcare systems, empowering healthcare professionals with actionable insights to proactively secure IoMT infrastructures. Evaluation of FOID on key IoMT cybersecurity datasets demonstrates its high efficacy, with overall accuracy rates of 98.53% and 97.91% achieved on the ICU and WUSTL-EHMS 2020 datasets, respectively, underscoring its potential as a reliable solution for anomaly detection in IoMT environments.