Our health secretary seems to be convinced that AI will help fix the NHS, and the UK government is also talking about AI in public services as if it’s some kind of magic wand that you can wave to solve every problem. This is not the case, and there are more important areas for attention.
In every system I’ve worked with – either as a developer or a user – there were daily frustrations that were either relatively easy to fix or changes that would give benefits far in excess of the effort required to implement them. Some examples from the NHS include:
Having to call for test results: Ringing up the GP surgery to find out whether your results have come in meant sitting in a queue for ages, only to be told that the results weren’t ready yet. Now they’re in the NHS app and you can check them immediately – although it’s not perfect as there’s no notification when new results come in, so you still have to poll for results.
Systems can’t exchange information: Our local hospital can’t see or update GP records, and our GP practice can’t see or update hospital records. This leads to frustrations for patients, who have to repeat everything they’ve told their GP, and wastes time when tests have to be repeated at the hospital so they’re on the right system (the alternative is for the hospital to ask the GP practice to post them the results and then scan them onto the hospital system). It’s also a safety issue if a patient is admitted as an emergency and the hospital can’t see their full medical history.
Switching systems: Our NHS trust is made up of 4 hospitals. Due to a pointless reorganisation (and rebranding), it’s ended up with 3 hospitals on one IT system and the other hospital on a different one. If you go to the consulting rooms, there are instructions telling you that you can’t be connected to both systems at once, and if you want to switch you have to call IT support.
Paper based blood test cards: If you need a blood test at our hospitals, there is an electronic system for requesting them, and everything shows up correctly when you arrive. However, this system is only available to certain staff, which doesn’t include everyone who needs to request or manage blood tests (e.g. nurses who arrange standard pre-surgery tests). If you don’t have access, you have to fill in a paper card with the tests required and post this to the patient, who has to bring the card with them to their appointment. This involves re-keying data and introduces a delay of several days whilst the paperwork goes through the postal system.
Batch prescriptions don’t show how many you have left: The NHS is moving to batch prescriptions for long-term conditions such as asthma. These are prescriptions where your GP pre-authorises a series of medicine which your pharmacy can dispense without any further approval, and saves the patient having to request a repeat dispense from the GP. However, it’s not possible to see how many dispenses are left in the NHS app – the only place a patient can get this information from is the printed prescription, which you don’t get to keep because everything is done electronically.
Unreliable systems: I’ve lost count of the number of times that the IT system for our GP practice has been down, which causes chaos for staff and patients. As a rough guess, there’s probably a 10% chance of something going wrong when I visit – this could be anything from the ‘next patient’ announcements not working to medical staff not being able to access records.
Only appointments can be added to the diary: I recently had a check-up with a nurse which involved some blood tests. She told that me that she would make an appointment for me in the calendar for a week’s time, but I should ignore it as it was to allow her to review the results and taken any follow-up action. If she didn’t put the appointment on the system, the time would be taken up with a ‘real’ appointment.
There are lots of other examples, and I know they frustrate members of staff because whenever anyone in the NHS finds out that I work in IT, they immediately and unprompted complain about the systems.
Don’t get me wrong, I’m excited about the possibilities for AI in some areas, such as analysing x-rays in conjunction with radiologists and taking advantage of the different things humans and computers bring to the table. But most systems have flaws that frustrate users on a daily basis, and fixing those will make users feel better and be more productive (fixing a small daily frustration is often better than a big annual frustration). AI is not a magic solution, and it is of zero help to basic problems like IT systems not being able to exchange data.
My advice to the UK government and anyone else? Fix the basics first, then think about whether AI can help with specific problems.