Story Highlight
- Report provides recommendations for safe AI deployment in NHS.
- Current regulatory framework deemed inadequate for healthcare AIs.
- Forty-four recommendations aimed at integrating AI into healthcare delivery.
- Patients should know when AI is used in their care.
- Human oversight is essential; AIs must not replace healthcare professionals.
Full Story
The National Commission into the Regulation of AI in Healthcare has published a report recommending the establishment of a comprehensive system for the safe implementation of AI tools within the NHS. Commissioned by the Medicines and Healthcare Products Regulatory Agency (MHRA), the report criticises the existing regulatory framework, which was developed for traditional medical devices, as inadequate for the complexities of AI in healthcare.
The report draws insights from more than 12,000 stakeholders, including patients, clinicians, and AI developers, and presents 44 recommendations aimed at integrating AI effectively into the UK’s healthcare delivery. Lawrence Tallon, chief executive of MHRA, stated, “We have heard clearly about the opportunities of AI to speed up and improve care, and to release more of clinicians’ time for the essentially human aspects of their work with patients.” He emphasised the necessity of a “modern, dynamic regulatory framework that accelerates safe adoption, protects patients, and commands public and professional confidence.”
Key suggestions include the introduction of staged authorisations for healthcare AIs, akin to learner driver L plates, allowing controlled testing prior to wider NHS deployment. The report also advocates for ongoing lifecycle monitoring of AI technologies instead of static approvals, facilitating the swift withdrawal of systems if they lose efficacy or safety. Authored by Prof Alastair Denniston from the University of Birmingham and Prof Henrietta Hughes, Patient Safety Commissioner for England, the recommendations urge updating liability laws to ensure developers face penalties for AI-related issues and improving the reporting system for adverse events linked to AI.
Patients should be informed about the role of AI in their treatment, with details regarding models accessible through a searchable public database. The report asserts that AI should supplement rather than replace human oversight, with final clinical decisions remaining the responsibility of healthcare professionals. “Any new innovation – including AI – needs to earn its place and earn our trust,” stated Prof Denniston, noting that public acceptance hinges on safety, human supervision, and transparency regarding AI usage.
The Health Services Safety Investigations Body (HSSIB), an independent body within the Department of Health and Social Care, expressed support for the report, highlighting the importance of balancing innovation with patient safety and public confidence. HSSIB has recently initiated an investigation into ambient voice technology (AVT) being deployed by the NHS, following concerns about potential safety issues and the adequacy of reporting systems for detecting errors associated with AI.
Interim chief executive of HSSIB, Dr Rosie Benneyworth, commented, “As this work progresses, we will carefully consider the Commission’s recommendations and the wider conversations taking place across the healthcare system about the safe use of AI,” adding that a better understanding of both opportunities and risks will assist in the safe integration of these technologies for the benefit of patients and healthcare staff alike.
Source: read the original report.
What this means for your site
The deployment of AI technologies in healthcare settings reveals significant gaps in the current regulatory framework. A shift from conventional approvals to staged authorisations, as recommended by the National Commission, can enhance safety during initial rollout phases. Implementing continuous lifecycle monitoring aligns with the principles of the Management of Health and Safety at Work Regulations 1999, ensuring proactive management of emerging risks associated with AI. The necessity for real-time oversight reflects the Health and Safety at Work etc. Act 1974, emphasizing the duty of care owed to patients by healthcare professionals.
A key area for improvement lies in establishing effective reporting systems; the concerns raise implications for RIDDOR 2013, highlighting the need for a mechanism that captures AI-related adverse events. Sites should prioritize training staff on identifying and reporting AI-related risks as part of their risk assessment processes. Additionally, creating a clear and accessible database regarding AI technologies in use not only complies with transparency mandates but also instills public confidence, ultimately enhancing patient safety. Proactively managing these aspects will mitigate potential harms and ensure that AI serves as a tool for enhancing, rather than replacing, the human element in patient care.
