AI: A plausible prospect or distant delusion in primary care?

Paul McNamara is a principal GP at the New Gorbals Health And Care Centre And honorary clinical lecturer at the university of Glasgow.

Nouf Aldhelaan is a 4th year medical student at the University of Glasgow

‘AI is set to revolutionize UK general practice by enhancing diagnostic accuracy and patient management. AI algorithms can predict health issues, enabling earlier interventions and personalized treatments, thus reducing chronic disease burdens. Additionally, AI streamlines administrative tasks like scheduling and record-keeping, allowing GPs more time for patient care. In diagnostics, AI excels in analysing medical images, detecting conditions like cancer with high precision. While the benefits are clear, addressing ethical concerns and ensuring data privacy is crucial. AI promises a future of predictive, personalized, and efficient healthcare in general practice’. (According to ChatGPT.)

That’s right, the previous paragraph was written by AI. Wowser! I asked ChatGPT to write me 100 words on the potential uses of AI in UK general practice, and it instantaneously churned out the above. I am simultaneously amazed and terrified!

Recent advancements in Artificial Intelligence (AI) have dramatically altered the functioning of various professions, offering innovative means of enhancing or even replacing human labour. The potential applications of AI throughout healthcare settings are a widely discussed topic with no shortage of proposed uses for AI in revolutionising the delivery of healthcare services. These range from AI-based radiological interpretations¹ to ‘virtual wards’ allowing patients to receive inpatient-level care at home, under the guidance of AI-based systems.² This is undoubtedly changing the future of medicine – but when does ‘the future’ become the present? Is AI currently an underutilised solution to the issues facing healthcare today, or does it remain an immature resource under development, only to be at the disposal of future generations?

AI has been paraded as a modern solution to the shortage of general practitioners within primary care. The Royal College of General Practitioners (RCGP) Artificial Intelligence and Primary Care report³ suggests a role for AI in proposing diagnostic tests and potential care plans to assist the human clinician in decision-making. Several research labs also claim AI may have equal or greater diagnostic accuracy, compared to its human counterparts.⁴ As promising or petrifying as these proposals may be, many challenges obstructing the implementation of AI in this context remain.

These challenges include safety concerns and costs, as well as clinician and patient impressions of AI. The most notable obstacle is concerns regarding patient safety and quality-control. One could argue AI may reduce human error directly by reducing human input and indirectly by reducing the workload on human labour. However, it is important to consider if we are simply exchanging human errors for systematic errors – and which of the two would be a more “acceptable” error to patients? Are patients more likely to forgive human error, thereby maintaining their confidence in the healthcare system? Another obstacle is the resources and development needed to create and implement these systems, as well as the technical staff needed to maintain them – which places a burden on the NHS’s already-strained budget. Lastly, it is of vital importance to consider clinicians’ and patients’ impressions of AI and the potential impact of AI on patient confidence. Overall, healthcare professionals are optimistic about the capabilities of AI. A survey conducted in Royal Free London NHS foundation Trust found that 79% of healthcare professionals deem AI to be useful or extremely useful.⁵ Another survey, conducted in China, assessing the response of medical staff to the concept of AI triage demonstrated a 77.1% acceptance rate and 45.2% preference rate for this.⁶ But where do the boundaries of patient comfort and confidence lie? Perhaps they would be comfortable attending an appointment following triage by an AI system, but would patients be comfortable taking a medication prescribed by an AI system, for a condition diagnosed by an AI system? While healthcare staff may be well-versed in the recent developments of AI and may be prepared to incorporate AI into their daily practice, the novelty of these systems may be a deterrent to some patients such as the elderly patient who had to mark their appointment on a paper calendar to ensure they don’t miss it. A survey assessing patients’ impressions of AI in healthcare found that only 55.4% of patients believe it could improve healthcare.⁷ demonstrating the threat to patient confidence posed by AI.

Realistically, any current applications for AI in primary care will be far less impressive than the extreme of AI-based consultations. Some smaller-scale roles for AI outlined by the RCGP include the upkeep and ordering of stock. AI may also be a useful adjunct in triaging patients toward the most appropriate healthcare professional for their respective presentation. The introduction of an interface between patients and AI systems also introduces patients to the incorporation of AI in primary care, in a manner that is more likely to be acceptable to them thereby preserving patient confidence. Nevertheless, some concerns remain irrespective of the grandeur of AI’s role within primary care – for example, the benefits offered by the adoption of these technological advancements by individual primary care practices may propagate the inverse care law, thereby increasing national healthcare inequalities.

Multiple roles for AI within primary care have been proposed, with many of the impressive roles being seemingly idealistic for various reasons. Nevertheless, the implementation of AI on a smaller scale to assist in primary care can be an important stepping stone to facilitate the implementation of large-scale AI-based healthcare in the future. This will allow clinicians and patients to gradually acclimate in this changing environment and pave the way toward the automated future of primary care.

Deputy Editor’s note: see also Richard Armitage on the many potential uses of AI here: https://bjgplife.com/using-ai-in-the-gp-consultation-present-and-future/ and Marcus Lewis’ reflections on using ChatGPT here https://wp.me/p9h5zY-5U6

References:

1. McKinney SM, Sieniek M, Godbole V, Godwin J, Antropova N, Ashrafian H, et al. International evaluation of an AI system for breast cancer screening. Nature. 2020; 577(7788): 89-94. doi: 10.1038/s41586-019-1799-6
2. National Health Service England. Virtual Wards. https://transform.england.nhs.uk/information-governance/guidance/virtual-wards/ [Accessed 18th June 2024].
3. Royal College of General Practitioners. Artificial Intelligence and Primary Care. 2019. https://elearning.rcgp.org.uk/pluginfile.php/174191/mod_book/chapter/504/artificial-intelligence-and-primary-care-jan-2019.pdf
4. Davenport T, Kalakota R. The potential for artificial intelligence in healthcare. Future Healthc J. 2019; 6(2): 94-98. doi: 10.7861/futurehosp.6-2-94
5. Castagno S, Khalifa M. Perceptions of Artificial Intelligence Among Healthcare Staff: A Qualitative Survey Study. Front Artif Intell. 2020; 3. doi: 10.3389/frai.2020.578983
6. Cao B, Huang S, Tang W. Ai triage or manual triage? Exploring medical staffs’ preference for AI triage in China. Patient Educ Couns. 2024; 119. doi: 10.1016/j.pec.2023.108076
7. Khullar D, Casalino LP, Qian Y, Lu Y, Krumholz HM, Aneja S. Perspectives of Patients About Artificial Intelligence in Health Care. JAMA Netw Open. 2022; 5(5). doi: 10.1001/jamanetworkopen.2022.10309

Featured photo by Steve Johnson on Unsplash.