How AI can improve productivity and efficiency from board to ward

AI adoption holds the promise of greater productivity for the NHS, but releasing this may be a challenge as NHS leaders battle with the day-to-day problems, as well as a lack of cash to invest. An HSJ webinar, in association with Multiverse, discussed what could be done to make this happen and where managers should be concentrating their efforts.

The type of technology we are talking about here is generative AI, said King’s Fund policy fellow Pritesh Mistry, which many people are now familiar with. Both the NHS and many local authorities have MS Copilot licences and are using it for activities like recording meetings, summarising those meetings and providing to-do notes, he pointed out.

“I would say we are really early days on a lot of the use of generative AI,” he said. Many of the pilots may have been using AI scribes or ambient voice technology, he said.

But is the NHS looking in the wrong place for adoption? Tim Horton, assistant director at The Health Foundation, said that when it surveyed clinical staff about AI’s potential last year, the responses focused on clinical documentation software, electronic patient records, and software for analysing images and test results.

This completely overlooked the real opportunities for using AI in administrative and operational work, he said, such as “the use of AI for triage, chatbot assistance for clinicians, the use of AI for booking, scheduling and communication around appointments, for rostering and so on and so forth… particularly if their process is like note taking, they’ll apply to a large amount of the NHS workforce.”

He added that, as such applications were “slightly further away from clinical decision making”, they might “carry less risk”, and therefore “there might be fewer regulatory hurdles around getting them in place [and it] might be easier to get staff buy-in”.

Starting with the problem

Mr Horton also warned “free trials and special offers”, rather than deep strategy, were currently driving some of AI’s adoption.

“We would certainly like to see the government and provider organisations thinking strategically about the most important applications and where attention and resources should be focused,” he said.

Organisations need to start with the problem they want to solve, added Rachael Watt, Multiverse account director, which would enable organisations to point everyone in the right direction.

“We know that organisations are so stretched, you cannot implement everything that everybody might want to in one go,” she said.

Multiverse data and AI coach Sara Richardson added: “There’s a lot of excitement among the NHS and eagerness to take on this technology. For me, what it comes back to is the need for a data culture and quality of data. You can’t implement these systems and these programmes without having that quality data in the right formats with the right purpose.”

Infrastructure and culture were also important, she said, so AI adoption could reach everyone in the organisation. “I will be talking to people on ward who are learning how to do these things, as well as people in their offices. So it is a huge range and it’s when you have the clinicians and the people that are doing the work every day that have that on the ground knowledge, this is where it really comes into its own, where they can use this rather than it being passed on to an IT department who doesn’t know what the reality looks like on the ground.”

One unanswered question was how the time released – and, in some instances, how the money saved – by AI is spent, she added.

Mr Horton said: “Staff have many calls on their marginal bit of time, which could be seeing more patients or research, but also management, admin, training. They might reduce their unpaid overtime. It follows then that actually using these to drive productivity improvement means them being part of broader work planning and part of broader change management.”

Mr Mistry added: “The technology is moving super fast, and our approaches to generating the evidence base do not move at pace compared to the speed of the technology.”

‘Making sure the building blocks are there’

But implementing AI means more than just adopting software. Mr Horton said: “It is to do with redesigning workflows, roles, work plans, getting buy-in from staff about how you’re going to use that freed-up time.”

Ms Watt added that organisations were looking for support around getting staff to think about how best to use these tools and gain the transformational benefits of AI. “You need to be able to upskill those folks who are facing the problems in their day-to-day,” she said. “It’s not just a kind of digital and data problem. This is something that impacts the whole workforce.”

What worked in one trust might not necessarily work in another, she added. “We find even accessing the data for a lot of apprentices that I work with is a challenge.” This could mean they were all asking the same person for assistance, she said.

AI adoption also needs to carry staff with it – and they may have concerns about what it means for jobs. Ms Watt said some staff would need support to link it to their organisation’s core goal.

She went on to say AI adoption should be treated as “a change management plan”, which involved “looking at the entire process and making sure the building blocks are there first to enable the next bits to happen”.

An on-demand version of this webinar is available.

Complete the registration form here, and you will then be sent details of how to access the recording.