Challenges Associated with the Adoption of Artificial Intelligence in Medical Device Software

Abstract

AbstractThe utilization of Artificial Intelligence (AI) has changed and enhanced several industries across the world, such as education, research, manufacturing and healthcare. The potential of AI to create new and enhanced applications that can benefit patients and physicians has created interest and enthusiasm, especially in a Medical Device Software (MDS) context. Although, the adoption of AI in MDS has also brought concerns for regulatory agencies and policymakers. The complexity of AI has challenged the standard requirements set by regulatory agencies, especially in the context of the differences between traditional MDS and AI. Additionally, the unique capacity of AI to continuous learning for optimal performance in real-world settings may also bring potential harm and risk to patients and physicians. The challenges discussed in this paper are in relation to: (1) Software Development Life Cycle (SDLC) frameworks; (2) learning processes and adaptability of AI algorithms; (3) explainability and traceability; and (4) conflictive terminology. At the end of this paper, conclusions and future work are presented to contribute to the safety and methodical implementation of AI in health care settings.