Debating the Challenges and Opportunities of AI in Healthcare
Exploring the Readiness of the NHS for the Clinical Implementation of Artificial Intelligence
Integrating Artificial Intelligence (AI) has always been a subject of discussion and debate. A recent dialogue hosted by the Digital Health Networks delved into the multifaceted perspectives on incorporating AI into healthcare, mainly focusing on its clinical use within the National Health Service (NHS).
Dr. Nisha Sharma, a prominent figure in breast screening and clinical lead for breast imaging at Leeds Teaching Hospital NHS Trust, and Dr. Marcus Baw, a locum GP and “General Hacktitioner”, articulated their belief in the utility of AI within clinical environments. However, they expressed concerns regarding certain segments of the health system being potentially unprepared for its use. They highlighted aspects of the technology that might not yet be sufficiently developed to be deemed reliable. Dr. Sharma, actively engaged in AI research, underscored the significance of distinguishing between low-risk and high-risk algorithms, particularly focusing on the latter, which are pivotal in making clinical decisions.
The challenges to integrating clinical AI are multifaceted and complex. Dr. Sharma elucidated several hurdles that complicate the seamless integration of AI into the health service. One such challenge is the lack of interoperability amongst different organisations, which she identified as a conspicuous factor. The necessity for substantial investment to ensure the utilisation of the most up-to-date software and to facilitate the accommodation of new algorithms was also highlighted. Furthermore, a shortage of expertise within NHS organisations was underscored with a call to action for quality assurance in addressing this issue. Understanding how the algorithms will be utilised and determining whether they will remain static or adapt as more data is accumulated was identified as a component in assessing readiness.
On the other hand, Dr. Baw spotlighted the absence of comprehensive guidelines for using AI in clinical settings. He noted that with experts globally advocating for enhanced regulation, it becomes exceedingly challenging to wholeheartedly support embedding AI tools into a system as crucial to our well-being as the NHS. He advocated for additional time and research to comprehend the potential risks associated with the clinical uses of AI. He highlighted the accountability of clinicians for decisions made using AI, as per the General Medical Council (GMC) and various legal bodies.
In contrast, Dr. Tracy O’Regan, from the Society and College of Radiographers, and Haris Shuaib, consultant physicist and head of clinical scientific computing at Guy’s and St. Thomas’s NHS Foundation Trust, presented a different perspective. They argued that AI, already embedded in parts of the NHS and part of routine care, is ready for clinical use in areas of clinical need. Dr. O’Regan emphasised that her profession has long-established standards set around new technology, including AI, that necessitate everyone working within it to comprehend what deep learning is and how it operates. She also highlighted that AI has existed in technologically advanced specialities such as radiography for several years.
Shuaib, aligning with this viewpoint, highlighted that there are already about 600 FDA-approved algorithms on the market, some of which have been available for over a decade. He also pointed out that approximately 300 CE-marked AI software products are available for clinical use. Shuaib acknowledged the controversies surrounding AI algorithms, such as instances where AI has disagreed with doctors’ opinions and cases where AI has identified suboptimal clinical performance.
The journey of AI in healthcare is undeniably complex, entwining technological advancements with ethical, practical, and procedural considerations. While initially, 54% supported the motion, by the end of the debate, those numbers were exactly reversed, with 46% voting against the integration of AI into clinical practice within the NHS.