The need for a holistic approach when deploying AI in clinical practice

Abstract

The deployment of AI systems in clinical care is increasing as new systems emerge and as trust in these systems develops over time. Healthcare is complex and whilst we embrace the notion of scientific breakthroughs in medical practice, clinicians hold an understandable level of caution about how systems will be reliable and aligned with human values and ethics, especially safety and patient autonomy. We propose a holistic approach to the implementation of AI with five guardrails designed to ensure that patient well-being, safety and privacy remain front of mind. The approach includes i) ethical foundations for safety and patient autonomy; ii) regulation such as that governing software as a medical device; iii) standards for providers encompassing guidelines, policies, procedures and governance structures for when systems are deployed; iv) the direct clinician-patient interface where doctors' duty of care and workforce education are paramount and finally v) consumer input and feedback which should guide decisions to deploy and form part of the adverse event reporting for the use of AI systems in healthcare. The need for shared accountability and ongoing monitoring is also discussed. Keyword: AI guardrails; medical duty of care; clinical governance; consumer empowerment, holistic AI deployment, workforce upskilling, agile safeguards.

Ähnliche Arbeiten