While I am generally a sucker for new technology, I do struggle with our increasing dependence on computers, particularly when dealing with people face to face. Whether you are buying screws at your local builder’s merchant, ordering a meal in a restaurant or consulting your GP, it’s likely that a computer/tablet/phone will get at least as much attention as you do. However, I have had a Damascene conversion this week. The future looks bright.
I doubt many of you will have read the ten-year plan to resurrect the struggling NHS in England but digital transformation, using technology and artificial intelligence (AI) to boost efficiency and improve patient access and care, is a recurring theme. And thanks to a new AI tool coming to a GP practice near you (if it’s not there already), I finally understand what all the fuss is about.
Ambient scribe may not sound that exciting but, trust me, you are going to like it. Put simply, with one click of a button (and your consent) your GP will be able to record all consultations. And when you have left, they press another button that converts the recording into clinical notes that can be checked before filing. And it doesn’t just transcribe what it has heard: it formats it into symptoms and clinical findings such as temperature and blood pressure, editing out all the irrelevant bits like the chat about the weather or your last holiday.
• Read more expert advice on healthy living, fitness and wellbeing
I can’t pretend to understand how it works but it is remarkable. You do have to change your consulting style a bit, as the computer can’t record what you are thinking so you have to explain everything to the patient as you go, which in itself isn’t a bad thing.
However, the biggest advantage is that you spend the whole time looking at each other — an important part of any consultation that has often been missing since computers landed on doctors’ desks. And it is not just GPs who are using the new technology, an increasing number of hospitals have made the switch too. Indeed AI is popping up everywhere in the NHS.
Radiology departments have been using computer-aided diagnostics for more than 30 years, starting with programs that can recognise worrying patterns in mammograms. And the latest smart tools are now being widely used to read x-rays and scans and help prioritise those that show worrying findings.
Like many specialities in the NHS, radiology is dealing with the double whammy of a shortage of specialists and fast increasing demand, and AI allows them to work faster and safer. Suffice to say that fewer cancers and other diseases are missed if x-rays and scans are checked by humans and AI.
Dermatologists have taken a step closer towards full autonomous AI, where the software alone decides which referrals need to be seen urgently. I send in a photo of a suspicious mole, the AI scans it and highlights possible cancers, and patients then get a face-to-face appointment with the specialist. And dermatologists need the help. Over the past ten years there has been a 170 per cent increase in the number of people referred to them for urgent assessment, just 6 per cent of whom end up with a diagnosis of skin cancer. Dealing with this huge demand in a timely manner, without missing a potentially lethal melanoma, is a huge challenge.
• Read more advice from Dr Mark Porter
Elsewhere in the hospital pathologists are increasingly using AI to help them detect cancers in biopsies taken by colleagues. At the moment this involves human and computer working together but given how accurate AI screening of biopsies now is, particularly in prostate cancer, how long will it be before humans only get to see the ones the AI has picked out?
Or in the case of the ambient scribe used in consulting rooms, how long before the AI starts making suggestions as to the diagnosis and management? Could this be Lyme disease, Dr Porter? Did you mean to prescribe that drug given the guidelines say it should be this one?
It’s not all roses. Some colleagues I have spoken to are concerned about the emphasis on AI diverting resources from other areas, not least adequate staffing. And, as evident in dermatology, it may exacerbate an already uneven playing field: about a third of hospital trusts in England don’t even have an urgent tele-consultation service for suspected cancers, let alone AI to help them interpret the photos sent in by GPs. Should resources not first be allocated to levelling up?
Finally, as with humans, it sometimes gets it wrong. Like the 82-year-old woman with a rash on her arm whom the new ambient scribe recorded as having “an itchy penis”. No, I can’t work that one out either, but that’s why we edit the notes before filing. For now, anyway …
AI in action
Here are three examples of AI software being used by the NHS
• In my part of the country (Gloucestershire) most GPs are using Tandem Health’s AI medical scribe. You can watch a demonstration at tandemhealth.ai
• Derm is helping to triage suspected skin cancers. See skin-analytics.com/ai-pathways
• And Ibex is an AI tool for pathologists looking at biopsies. See ibex-ai.com