With the pandemic shifting views on AI and data use across industries, Maja Dragovic investigates how much those views have been altered in the NHS.

Necessity, as the saying goes, is the mother of invention and during Covid necessity was fundamental in adopting inventions already available. In particular, the use of AI and data and the tools that were readily available, were embraced with more readiness than they might have been greeted before.

“There was no alternative,” says Ben Court, head of analytics at Civica. “We had to use some form of data to make decisions, and we had a clear list of problems, and we had access to technologies that allowed us to do it.”

As Court points out, often with new technologies, be that business analytics, raw data and reporting, AI or machine learning, people can focus on technology and what it does rather than the problem that needs to be solved. But this rule does not apply in healthcare, especially during a pandemic.

“Our experience certainly is that good AI solutions or good data solutions derive from a problem,” Court says.

“There are different ways of developing a data solution and if you focus on the problem, and not the technology, you should always be tailoring it to the individual and what they need.”

Time pressures

In the middle of the pandemic, time was of the essence. There was no time for manual analysis, and quick access to up-to-date comprehensive information was crucial to decision making, especially in the ever-changing situation.

When it comes to AI, Court explains, it was speed and detail that proved to be invaluable during the pandemic.

“Across the hospital you can have 60 to 100 specialities and under those specialties you may have a dozen services each, and if each of them needs to forecast the impact on beds or the impact on medicines, you need something that can do that very quickly and in a real time fashion because the situation was changing on a day by day basis,” he adds.

“It was the enterprise level intelligence that AI can provide that was the real value within the pandemic.”

Automation of administrative tasks

Tom Hockton, BMC’s regional sales director, points out that many hospitals started leveraging AI to help reduce time on administrative tasks. A number of trusts started to roll out chatbot technology, effectively powered by AI, to be able to act as a new channel of communication.

Rather than having staff working on simple and repetitive tasks, Hockton says a chatbot was leveraged to handle a lot of the requests that come in such as “I need a new laptop” or “I want to request some time off”.

Another example is checks on oxygen levels that is usually done manually, up to four times a day. This can be done by a bot for practically zero cost, says Gary Pruden, chief technology officer at Fusion Global Business Solutions, and they can do it every minute.

“The accuracy increases, and it also takes away the need for someone to check it manually,” Pruden adds.

He also notes that the reason for increased interest in automating tasks is a consequence of having a number of staff being forced to work remotely during the pandemic.

“There was a mad scramble at the beginning of the pandemic to allow people to work remotely,” Pruden says.

“(But) I think most organisations found that some data was not available remotely.

“People have realised that we can’t just have people going into the office and tweaking things all the time, it’s inefficient and during the pandemic it’s also quite dangerous.”

He adds, however, that due to the sheer number of automation possibilities, there is almost “analysis paralysis” with many organisations not knowing where to start. The issues are usually around whether to automate the complete processes or individual parts.

Data sharing uptake

The pandemic has also accelerated the sharing of data between different care providers and bringing data together from different systems.

Linking data from social care is one example, says Richard Betteridge, data scientist at Cerner. He notes that this is possibly due to the health and care inequalities that have been exposed by the pandemic.

“A lot of information is well contained within social care records and it isn’t necessarily picked up by healthcare,” Betteridge explains.

“So, having that linkage from health to social care data, and vice versa, really helped enable some of the additional work for quantifying the impact of some of those social determinants components.”

His colleague, Charlie Evans, analytics team manager at Cerner, believes that data from wearables is coming into the fore and is going to play a huge role in the future. As an example, he points to “a big push during covid to get the information from oximetry meters fed back (into the patient record)”.

Continuing to deliver

As the benefits of AI systems in healthcare start to be realised, Court believes that this momentum has to continue and points to the recent past where lessons can be drawn.

“There was a real threat in late 2019, early 2020 that we were going to go into the third AI winter,” he says.

“The lack of deliverables was starting to come through. And then we had the pandemic, which was always a bit of a saving grace for AI in that era because it started delivering, it gained a focus on a problem rather than being about the hype.”

As we come out of the pandemic and if the hype train kicks off again, Court adds: “We will have a lot of people not investing in the right ways, not delivering systems in the right way and then we can end up in that bust again”.