AI has now hit the mainstream and is transforming many industries. Not a day goes by without a new story on AI’s potential in healthcare. However, setting aside the hype, it’s worth taking stock of where we are and considering the direction of travel to implement AI into routine clinical practice at scale in the NHS, a system which is typically slow to implement new IT technologies.
Many of the requirements will be the same as for any digital solution. IT system integration is an obvious one for a technology powered by data – and NHS Scotland has a real advantage here – but it also needs to fit into the clinical pathway. Information governance, security and compliance challenges will apply, as always, with the added complication that the technology is poorly understood by regulators.
However, the key thing for successful implementation is to start with the problem and not the technology. AI is a tool in the toolbox, not a magic wand to solve all the problems of the NHS, and certainly not something which can be deployed in isolation.
Also, most current AI initiatives are funded through research and innovation. These demonstrate the ‘art of the possible’ but it’s important to remember the chasm between R&I projects and the successful adoption, funding and implementation of these projects into routine clinical practice.
Realistically, in the current financially constrained environment, AI solutions will need to save NHS time or improve patient outcomes without increasing clinical workload. The £150 million AI framework recently announced in England may change this but there is currently no equivalent funding in Scotland to move more ambitious AI technologies (e.g. early detection of cancer) from pilots into practice.
Additionally, if your clinical service currently relies on paper and spreadsheets, there may be a prior step of digitisation before you can truly reap the benefits of AI.
A couple of use cases from Red Star illustrate the approaches. In one, we use AI to read through the clinical records of thousands of heart failure patients and then proactively identify those whose treatment can be optimised. This essentially ‘beats the bureaucracy’ to improve patient outcomes and can do in a few hours what it would take clinicians months to accomplish.
In another, we use AI to transcribe the paper forms of patients on a waiting list for ADHD. In itself, this replaces a time-consuming admin process but also digitises the backlog allowing for further downstream improvements in efficiency and the quality of care delivered to patients.
Radiology is an area of real promise due to the maturity of computer vision. And the rise of ChatGPT will see a lot more text-based use cases – needed in healthcare – however, there are significant technical and governance challenges in applying large language models to patient data.
In all likelihood, the impact of AI may be over-estimated in the short term and under-estimated in the medium and longer term. The real strength that Scotland can capitalise on are the IT systems integrations already in place through SCI-Store, NSS and NDP, allowing for a Once for Scotland approach and overcoming what will be a significant barrier elsewhere.
All the building blocks are there and it’s an exciting time to be working in AI. Hopefully, with the right tailwind from the NHS and Scottish Government, we can ensure the benefits are delivered sooner rather than later.
Red Star will be delivering a masterclass on ‘How to Successfully Implement Digital Solutions in the NHS‘ at Health & Social Care Transformation next Tuesday in Glasgow. Register HERE
