Abstract:Internet of Medical Things (IoMT) applications, called also Medical Internet of Things (MIoT) or IoT for e-health, allow integrating smart technologies to medical devices for a better monitoring of disease progression and patients tracking. While it offers extensive benefits and despite of its expansion, IoMT brings round additional security concerns. Connected Medical Devices (CMD) raise up several security and privacy problems. The complexity and heterogeneity of data and technology in IoMT communications create additional security risks and threats. Malicious users may exploit crucial vulnerabilities in a wide range of IoMT applications, networks and devices. Hence, thinking about smart and efficient security solutions is an urgent need to understand and assess IoMT-related threats. Existing traditional models are no longer convenient and remain unsuitable to address the various born risks. We present in the current manuscript an in depth study of security concerns within IoMT. We review popular risk assessment and management approaches and discuss their suitability to the IoMT context. The main shortcomings are inherent to the complex architecture, the lack of automation and the numerous stakeholders with different security needs and skills.With reference to the conducted study, we introduce our solution as a framework to enhance trustworthiness and support decision making within IoMT environments. The proposal relies on a fine-grained approach for managing associated risks with regard to different areas of focus and common risk factors. To do so, a Machine Learning (ML)-based anomaly detection model and a hybrid Risk Assessment (RA) model are defined to evaluate cumulative IoMT risk. Experiments show obtained competitive results compared to state of the art ML models to detect intrusions in IoT/IoMT systems and we obtained an accuracy rate of 100% with some algorithms. A use case of application is presented to highlight the efficiency of the proposal.

MLSMM: Machine Learning Security Maturity Model

Security for Machine Learning-based Software Systems: a survey of threats, practices and challenges

Security and Machine Learning in the Real World

Characterizing machine learning process: A maturity framework

RMF: A Risk Measurement Framework for Machine Learning Models

Towards the Science of Security and Privacy in Machine Learning

Risk Management Framework for Machine Learning Security

ML-On-Rails: Safeguarding Machine Learning Models in Software Systems A Case Study

Usable Security for ML Systems in Mental Health: A Framework

SecurityNet: Assessing Machine Learning Vulnerabilities on Public Models

MLRA-Sec: an adaptive and intelligent cyber-security-assessment model for internet of medical things (IoMT)

Mental Models of Adversarial Machine Learning

SoK: Applying Machine Learning in Security - A Survey

The Role of Machine Learning in Cybersecurity

Toward a Research Software Security Maturity Model

RM4ML: requirements model for machine learning-enabled software systems

Adversarial Machine Learning -- Industry Perspectives

The Risks of Machine Learning Systems

Beyond the ML Model: Applying Safety Engineering Frameworks to Text-to-Image Development

Technology Readiness Levels for Machine Learning Systems

A Marauder's Map of Security and Privacy in Machine Learning