Enhancing Cybersecurity in Healthcare: Evaluating Ensemble Learning Models for Intrusion Detection in the Internet of...

Area of Science:

• Cybersecurity
• Machine Learning
• Internet of Medical Things

Background:

• The Internet of Medical Things (IoMT) is expanding, increasing the attack surface for sensitive healthcare data.
• Robust cybersecurity measures are crucial to protect patient privacy and ensure the integrity of medical devices.

Purpose of the Study:

• To evaluate the effectiveness of ensemble machine learning algorithms for intrusion detection in IoMT environments.
• To compare the performance of Stacking, Bagging, and Boosting ensemble methods in identifying cyber threats.

Main Methods:

• Utilized the WUSTL-EHMS-2020 dataset for training and testing intrusion detection models.
• Implemented ensemble learning techniques: Stacking, Bagging, and Boosting.
• Employed Random Forest and Support Vector Machines as base models within the ensemble framework.

Main Results:

• Stacking ensemble learning achieved the highest accuracy at 98.88% in detecting and classifying cyber attacks.
• Bagging ensemble learning demonstrated strong performance with an accuracy rate of 97.83%.
• Boosting ensemble learning yielded a lower accuracy rate of 88.68% compared to Stacking and Bagging.

Conclusions:

• Ensemble machine learning, particularly the Stacking approach, offers a highly effective solution for intrusion detection in IoMT.
• The findings highlight the potential of advanced machine learning techniques to bolster cybersecurity defenses in the healthcare sector.