Artificial Intelligence (AI), like all other technologies used within the NHS, can be a tool to increase productivity, improve outcomes and deliver better patient care. Most, if not all NHS board members and senior leaders have been through the implementation of an electronic patient record (EPR) or a similar large scale transformation programme, and have scars to bear as well as achievements to celebrate.
As the focus on AI increases, from the prime minister's "blueprint to turbocharge" and the government's "plan to leverage technology", and with it likely to only increase further with the 10 year plan, there are key learnings that board members and senior leaders can take from experiences of EPR implementations, to be applied when thinking about AI. These can help boards in their critical role of leading successful transformation, realising the expected benefits and ensuring better outcomes for staff, patients and the wider NHS.
But before we dive into that, we should start by saying that AI in the NHS is not new. In fact, AI in different forms has been around in the NHS for a number of years - machine learning is helping to improve back office efficiency and deep learning is helping clinicians review medical imaging. However, it's the more recent explosion of Large Language Models (LLMs) like ChatGPT which are very large deep learning models that are pre-trained on vast amounts of data, which have really caught the attention. This is why AI is a topic on the agenda of most trust boards right now. If any of the terms used in this report are a little puzzling, the NHS AI Dictionary is a fantastic resource to help - you can access it here.
We're seeing and hearing of well considered, successful and impactful use of AI in the NHS. For example:
- The COLO-DETECT study, led by South Tyneside and Sunderland NHS Foundation Trust and Newcastle University has shown an 8.3% increase in detection of cancer cells at Bolton NHS Foundation Trust which they estimate could lead to a 42% reduction in fatal cases.
- At South Tees Hospital NHS Foundation Trust, predictive AI has helped NHS teams identify those at risk of becoming High Intensity Users (HIU), enabling dedicated key workers to provide proactive support, reducing the number of HIU A&E visits by half.
- Trusts and system partners in Lancashire and South Cumbria Integrated Care System (ICS) are using an AI technology, DERM, to create patient pathways to support faster skin cancer diagnosis. Across the ICS, 54% of potential urgent suspected skin cancer referrals have been avoided and £168,000 worth of GP appointments have been reallocated to other patients.
In fact, the health sector does have enormous potential for machine learning techniques as there is a significant amount of very large, very well structured, and very well labelled datasets to train the models on, with radiology imaging being the foremost example can be seen here.
In this report, we share lessons learned from previous technology implementations in the NHS and questions board members can use to assure themselves about AI, not to play down or diminish the opportunities, but instead to help think about what we can do to better set up for success and enhance what is already available.