- Stephen H Bradley, GP and NIHR academic clinical lecturer12
Patients don’t need AI doctors, but better technology could allow medics to spend more time caring for patients
Much of the excitement around using artificial intelligence (AI) in healthcare lies in the promise that machines will be able to make sophisticated clinical judgments, such as diagnosing disease and making treatment decisions. In the past year the UK government has invested in AI in the hope that it can speed up diagnosis and tackle NHS waiting times. The former health and social care secretary Steve Barclay boasted that “more cutting edge AI technology across the NHS” will help with “quicker, more accurate diagnosis of lung cancer.”1 Similar hopes have been echoed by other healthcare leaders and commentators.2
AI will undoubtedly revolutionise radiology interpretation, even if the evidence supporting plans to procure the technology for English hospitals is currently lacking.23 Other applications touted for AI include acting as “virtual nurses” for patients with chronic diseases and predicting cancer metastases.4 Such technology may well prove useful in time, but the most exciting and plausible near term promise of AI isn’t that it could replace doctors with computers—it’s that doctors might be freed from the drudgery imposed by our current computer systems.
People who rhapsodise about AI driven healthcare often seem to have little appreciation of how the functionality and user experience of our present systems lags decades behind consumer technology that we take for granted. Systems that are notionally paperless are actually a mishmash of digital and analogue, with vital information trapped in non-searchable PDFs and scanned letters. For example, GPs receive discharge letters electronically, but in my case this hasn’t made the task of reconciling medicines any easier and usually requires me to print off a copy to tally the changes manually, one by one. Finding out a patient’s diagnosis or working out why a hazardous medication was started often requires slowly loading different pages of scanned detritus before finding the one document that holds the explanation.
An NHS with an IT infrastructure that would “work” in the ways we wanted could transform clinicians’ working lives and bring some joy back to the job, allowing doctors to stop squinting at screens and spend our time talking to patients instead. In primary care many GPs have highlighted the administrative burden they contend with, and in one survey of GP trainees they cited this “hidden” workload as one of the top reasons they wouldn’t want to work full time.5 These administrative tasks are made considerably worse for doctors by having to use systems that aren’t fit for purpose and leave us struggling with innumerable fiddly annoyances that we have to fit around patient care.
Breakthroughs within reach
We could vastly improve the efficiency of our healthcare system, and save healthcare staff and patients a great deal of time and frustration, if we had systems that could automatically:
Answer questions such as, “Who started this drug and why?”
Identify outstanding care and reviews and then contact patients to book them in
Allow staff in general practice to reconcile drugs started in hospital or vice versa, rather than relying on manual, error prone transcription; and
Field requests to provide evidence for reports, including for benefit claims, insurance companies, or coroners’ inquiries, complete with third party redactions.
It would also be helpful to have tools that could guide doctors and patients through complicated chronic disease guidelines, enabling us to start every patient encounter with a problem list drawn from clinic letters and free text entries, along with a summary of recent healthcare encounters and relevant investigations.
To many policy makers this vision may well seem utterly unambitious. Those who don’t work in healthcare would understandably assume that our systems already talk to each other and do many of these things. But focusing on the breakthroughs within reach that could bring transformative benefits to quality, safety, and staff wellbeing is a much sounder proposition than indulging in farfetched fantasies that remain entirely remote from the ramshackle patchwork of our existing health IT.
Doctors don’t need to worry about being replaced by machines any time soon. Even if AI could master diagnosis and the interpretation of tests, these tasks constitute only a minority of what doctors do that brings value. Listening, explaining, weighing up options with patients, and gauging the best fit with their preferences and values is still the foundation of high quality healthcare.
Instead of dreaming of creating ersatz doctors, our AI developers and policy makers should concentrate on technology that frees clinicians from the drudgery of data entry and retrieval, freeing us to do more of the fulfilling human tasks that we’re best at and those that patients need the most. If we had systems that “worked” we still wouldn’t have enough time, but we could at least spend the time we do have serving our patients, not the electronic health record.
Footnotes
Competing interests: SHB is clinical lead for cancer for the Leeds office of the West Yorkshire Integrated Care Board.
Provenance and peer review: Not commissioned; not externally peer reviewed.
References
- ↵
Barclay S. Department of Health and Social Care. AI to speed up lung cancer diagnosis deployed in NHS hospitals. Gov.uk. 30 Oct 2023. https://www.gov.uk/government/news/ai-to-speed-up-lung-cancer-diagnosis-deployed-in-nhs-hospitals
- ↵
Barclay S, Smith C, Department of Health and Social Care. £21 million to roll out artificial intelligence across the NHS. Gov.uk. 23 Jun 2023. https://www.gov.uk/government/news/21-million-to-roll-out-artificial-intelligence-across-the-nhs
- ↵
National Institute for Health and Care Excellence. Artificial intelligence-derived software to analyse chest X-rays for suspected lung cancer in primary care referrals: early value assessment. Health technology evaluation. 28 Sep 2023. https://www.nice.org.uk/guidance/hte12
- ↵
- Webster P
. Six ways large language models are changing healthcare. Nat Med2023;29:2969-71. doi:10.1038/s41591-023-02700-1.pmid:38036704
OpenUrlCrossRefPubMed
- ↵
- Khan N
. “Hidden” no more: the real work of the “part-time” GPs working full-time hours. Br J Gen Pract2023;73:26-7. doi:10.3399/bjgp23X731625.pmid:36543544
OpenUrlFREE Full Text
