Today I’ve been ‘acting down’ to cover the registrar rota for the strikes. All OK so far, I’m just hoping not to get called out too much tonight! Over the last month I have written about what in my view have been the biggest advances in heart health in the last 10 years. This has lead me to reflect, what will the biggest advances be in the next 10?
I recently read an article in the medical press by a doctor concerned with patients coming to her with an AI generated diagnosis simply wanting the doctor to run a test or prescribe a treatment. I’m sure if I’m still writing my blog in 10 years, the biggest change will have been in AI, but in 2025, my job is still safe! Whilst AI is great at generating very sensible sounding lists of potential diagnoses, and what the tests and treatments are for those, it is still some way from replacing a good doctor in many areas of medicine. For the time being at least (and I hope forever) there will always be people who would prefer to talk to a human than an AI. In addition, whilst an AI does sound very confident when it produces those lists, there are often mistakes which it simply can’t spot, and neither can the patient.
Certain areas of medicine however lend themselves very well to AI. I could easily believe it won’t take long until x-rays, CTs and MRIs are all reported by an AI. If I get a skin issue, photographing it and getting an AI to tell me what the rash or mole is seems a great option instead of a long wait. Surely soon I won’t get emails from GPs and other non-heart specialists asking me to interpret an ECG for them? It won’t be long before an AI is better than me at that.
How about counselling or psychiatry? A major barrier to effective psychotherapy is a poor therapist-patient relationship. Finding another human you gel with might be tricky, maybe there are no others in your area, or maybe they are too expensive? No such problem with an AI – simply change them!
Many surgical procedures already use computers in one form or another to get better outcomes. We use them in the cath lab already expecially in ablation procedures, and robotic surgery for some cancers and orthopaedic operations is commonplace. However, I think the government’s assertion that AI surgery will bring waiting lists down is a fallacy, at least during my career. Robotic procedures often take longer (for now) than their human counterparts, but are more accurate given the precision inherent in a robot. That may be great for surgical success rates, but not so good for waiting lists.
I read an excellent article recently about productivity in the NHS. Productivity is a key media buzzword in many areas of modern life. We often hear that despite record investment in the NHS, we are less productive as an organisation. The article in question compared the number of steps in the surgical pathway needed to complete an operation (in this case a tonsillectomy) in 2025 compared to 2005. The actual operation time has remained remarkably constant (I guess tonsils look the same as they did 20 years ago, and so does a scalpel). However, the number of steps (the bulk of which are administrative) has increased more than 500% (from 46 to 265)! I’m sure many of us feel we are over-burdened with admin, I certainly am. However whilst at first glance it is easy to feel frustrated by how complex a tonsillectomy has become, a deeper dive reveals how many of these steps are to do with safety. If we only measure productivity in terms of (in this case) the number of operations completed, no wonder it appears that we have zero more productivity to show for all that investment. However, if one of my kids needs their tonsils out, I want to be sure it is as safe as possible. There is lots of data demonstrating that surgical procedures in 2025 are safer than in 2005. We therefore need a way of incorporating these complexities into the NHS productivity statistics, as a good health service should absolutely be safe as well as productive. Maybe ChatGPT can help?
A huge obstacle for AI will be indemnity. Who do you complain to if your AI doctor has made a mistake, the algorithm? Who ends up in court? I’m sure there will be a way through all this, but I do somewhat fear for doctors of the future. Their careers are likely to be rather different from my own. I was always guaranteed a job during my training, but this year 30,000 doctors in training were competing for 10,000 jobs. I’m glad I’m a consultant.
In this week’s research video, a trial that was music to my ears. I am a coffee lover, and self-confessed coffee snob (I would sooner have no coffee than bad coffee). I would even rather never drink alcohol again than never drink coffee again. I was delighted therefore to read about a trial presented at the American Heart Association meeting in New Orleans last week looking at coffee drinking and atrial fibrillation. Do check out the video!