[Paper Review] Explainable Machine Learning-Based Security and Privacy Protection Framework for Internet of Medical Things Systems

Ayoub Si-Ahmed, Mohammed Ali Al-Garadi+1 more|arXiv (Cornell University)|Mar 14, 2024

Artificial Intelligence in Healthcare and Education5 citations

TL;DR

The paper presents a federated learning–based intrusion detection framework for IoMT that uses ANN for anomaly detection and XAI techniques to explain model decisions, aiming to preserve privacy and achieve performance comparable to centralized methods.

ABSTRACT

The Internet of Medical Things transcends traditional medical boundaries, enabling a transition from reactive treatment to proactive prevention. This innovative method revolutionizes healthcare by facilitating early disease detection and tailored care, particularly in chronic disease management, where IoMT automates treatments based on real-time health data collection. Nonetheless, its benefits are countered by significant security challenges that endanger the lives of its users due to the sensitivity and value of the processed data, thereby attracting malicious interests. Moreover, the utilization of wireless communication for data transmission exposes medical data to interception and tampering by cybercriminals. Additionally, anomalies may arise due to human error, network interference, or hardware malfunctions. In this context, anomaly detection based on Machine Learning (ML) is an interesting solution, but it comes up against obstacles in terms of explicability and privacy protection. To address these challenges, a new framework for Intrusion Detection Systems is introduced, leveraging Artificial Neural Networks for intrusion detection while utilizing Federated Learning for privacy preservation. Additionally, eXplainable Artificial Intelligence methods are incorporated to enhance model explanation and interpretation. The efficacy of the proposed framework is evaluated and compared with centralized approaches using multiple datasets containing network and medical data, simulating various attack types impacting the confidentiality, integrity, and availability of medical and physiological data. The results offer compelling evidence that the FL method performs comparably to the centralized method, demonstrating high performance. Additionally, it affords the dual advantage of safeguarding privacy and providing model explanation while adhering to ethical principles.

Motivation & Objective

• Motivate secure IoMT deployments by addressing security and privacy risks in wireless data transmission and anomalies in medical data.
• Propose an ML-based IDS architecture that preserves privacy via Federated Learning and avoids centralized data collection.
• Integrate XAI methods to enhance explainability and trust among patients, designers, and regulators.
• Evaluate the proposed framework against centralized approaches on multiple network and medical datasets.

Proposed method

• Design an IDS framework for IoMT using Federated Learning to train models on local devices and share model weights rather than raw data.
• Implement deep learning–based intrusion detection with Artificial Neural Networks for real-time anomaly detection.
• Incorporate XAI techniques to explain and interpret model decisions for stakeholders.
• Use a three-layer IoMT architecture (Data Acquisition Layer, Personal Server Layer, Medical Server Layer) and optimize communication rounds and local epochs in FL.
• Provide a detailed training setup with Adam optimizer and binary cross-entropy loss for local model updates.
• Compare FL with centralized learning across four datasets to assess performance, privacy, and explainability advantages.

Experimental results

Research questions

• RQ1 Can Federated Learning achieve comparable intrusion detection performance to centralized training in IoMT environments?
• RQ2 Does incorporating XAI provide meaningful explanations that enhance trust without compromising detection accuracy?
• RQ3 What are the privacy, latency, and scalability implications of FL-based IDS in IoMT across multiple datasets?
• RQ4 How does the proposed ANN-based IDS perform on network and medical data combinations in IoMT?

Key findings

• FL-based IDS achieves high performance comparable to centralized methods across evaluated datasets.
• Federated Learning preserves privacy by exchanging model weights instead of raw data, reducing bandwidth and latency concerns.
• XAI integration improves interpretability for patients, designers, and regulators without sacrificing detection capability.
• Experiments demonstrate the framework’s applicability and robustness across four distinct datasets containing network and medical data.
• The proposed framework addresses single-point-of-failure risks associated with centralized models and supports system scalability.