On the cusp: early insights on non-clinical AI in the NHS

Is non-clinical AI one of the answers to the NHS’s challenges? 

Many attendees at March’s ReWired digital health conference were abuzz with the potential of artificial intelligence (AI) to make work in health care more enjoyable and lessen the risk of burnout. For an NHS workforce experiencing high levels of stress and more vacancies than can be filled, this is an attractive proposition. 

While the huge burden of administrative tasks is just one of the challenges the NHS faces, there’s increasing enthusiasm about the potential of AI to improve back office and non-clinical processes in the NHS – not only boosting productivity but improving working conditions too.   

Office for National Statistics figures on public sector productivity highlight the urgent need to address the NHS’s productivity, which took a hit during the COVID-19 pandemic and is yet to recover to pre-pandemic levels. The government’s recent AI Opportunities Action Plan sees AI as a key enabler for improving health care and productivity across the economy.  

From sending prompts to the patients most likely to miss an appointment to enabling more efficient surgery scheduling, and transcribing and organising the salient points from a patient consultation, AI could make a range of processes more efficient. Our research exploring attitudes towards AI and technology also suggests NHS staff and the public are on balance supportive of the use of AI for administrative purposes (with 61% of the public and 81% of NHS staff surveyed in favour). 

Yet there’s still much more to learn about which AI applications are most promising to address NHS challenges For example, our recent research highlights that technologies like AI have the potential to free up time for staff, but it depends how they are implemented and used. If the NHS is to unlock the benefits of AI, we need better information about which use cases have the greatest potential and how to implement these effectively and responsibly.   

It’s against this backdrop the Health Foundation is exploring how non-clinical AI can be used and spread responsibly in the NHS.  

The current state of non-clinical AI in the NHS 

Limited awareness of use cases  

Last year, we asked stakeholders from across the NHS – including chief operating officers, chief nurses, partners in GP practices and others – which non-clinical AI use cases they know about and what challenges they are trying to address, as well as about their experiences using non-clinical AI. This gave us a range of insights. Leaders we spoke with were generally not very familiar with examples of AI applied to non-clinical tasks and were waiting to learn from others ahead of adopting it.  

There is uncertainty about where AI could have the greatest impact, making it hard to know where to invest, and financial constraints are limiting the ability of some NHS organisations to procure AI technologies. Stakeholders told us about the need to get the basic data and digital infrastructure right to integrate AI products and the difficulties in securing the skills and time to implement AI well, underlining the importance of supporting the change as well as the tech. 

Decision makers want to know what solutions are right for their needs  

We also heard of the groundswell of products being marketed to the NHS, sometimes without sufficient information for leaders to make informed decisions about what to procure. We heard that many solutions offered to NHS buyers lack the information or evidence decision makers need to have confidence the products will deliver the results they are looking for. The NHS stakeholders we spoke with told us they don’t always feel they have the knowledge to make informed decisions about the effectiveness or value for money of these AI solutions. We also know there is uncertainty about which use cases could be classified as medical devices, which can be hard for buyers to navigate given the different approval and use requirements associated with medical and non-medical devices.  

These insights chime with findings from our recent research with UCL Partners to explore the state of AI adoption in London. While some of the challenges are common to health technology more generally, the particular complexity and risks associated with AI can make it more difficult to make progress with adoption. 

A burgeoning market of non-clinical AI solutions 

To explore the area further, we commissioned the https://healthinnovationnetwork.com/Health Innovation Network South London to build a snapshot of the non-clinical AI market, existing use cases and the difference these use cases are making for NHS organisations.  

This snapshot of the market found 190 examples of non-clinical AI products being marketed to the NHS across a range of uses in corporate, operations and clinical operations categories. We found the greatest current use of non-clinical AI in HR and training, booking and scheduling, staff efficiency (including ambient voice technology), staff rostering and medicines management. While this is a rapidly moving market, and it’s not always immediately clear if AI is part of a product or not, this gives us a better sense of what is currently available.  

Following this market review, NHS organisations across the Health Innovation Network were asked which of the non-clinical examples were being used. The responses revealed 46 examples of non-clinical AI being implemented across 70 different NHS sites in England. This tells us the same product is sometimes being used at more than one site, and some sites use more than one non-clinical AI product. The spread of these examples extends across England, with a disproportionate concentration in London. 

The demand for booking and scheduling tools and staff efficiency and rostering products is particularly noteworthy, given these tools could be used across most parts of the health care system.  

Case studies will shed light on implementation and indicative impact  

This work is just the start of a bigger programme at the Health Foundation to support the responsible use and spread of non-clinical AI.  

In the summer, we will publish case studies sharing the experiences of NHS organisations currently deploying non-clinical AI, alongside implications for practitioners and policymakers. 

These case studies will provide insight on the implementation of non-clinical AI solutions, patient and staff involvement and the impact seen. Will these products help improve work in health care? Stay tuned to find out. Register to get updates about our work on technology in health care.