Big Data, analytics and technology - opportunities for encouraging innovation and adoption
Guy Boersma, MD of KSS AHSN, spoke at the Westminster Health Forum's debate on the future of digital technology in the NHS.
The transcript of Guy's keynote speech can be read below. A transcript of the full event is available from Westminster Health Forum.
Understanding artificial intelligence and its adoption in the NHS
I thought I'd just start with that joint report that was published last week that talked about the need for technology to contribute more to health and social care now and in the future. I'm talking about that joint report from the Health Foundation, the King's Fund and the Nuffield Trust about the current state of the workforce challenges in the NHS and in social care. I thought I'd just see if other people had spotted what it was saying and start a couple of questions to you about the size of the NHS workforce across the UK: about how many people are employed in the NHS these days?
Chris Ruane MP: One and a half million.
Pretty good, 1.2 million. And in social care in addition to those 1.2 million of NHS staff about how many people are employed in social care?
Chris Ruane MP: Two.
Two combined, so 1.2 in the NHS, 1.1 in social care, so about 2.3 million. And those 2.3 million staff represent about 10% of the UK workforce. With the demographics of a growing and ageing population it's forecast by the Office of National Statistics that by 2030 there are going to be 36% more folk aged 65 and over, so, what used to be called retirement age group and only 3% more 18-65-year-olds. The challenges for getting staff to work in the NHS with a growing caseload are only going to get tougher. And what was reported last week in the joint report was the size of the workforce challenges currently faced. Did anyone pick up that of the 1.2 million workers in the NHS, how many more vacancies there were across the NHS at the moment in addition to those 1.2 million?
Chris Ruane MP: 100,000.
Another 100,000, yes, and the same in social care, so sort of 200,000 or so vacancies and it's hard to see where those vacancies are going to be filled tomorrow, let alone in 2030 when the demographics are against the working age population.
So final two things, one how many nursing posts are vacant at the moment? 1 in 8 nursing posts are vacant at the moment.
Chris Ruane MP: And there are more nurses not nursing than there are nursing, they're retired, or resigned, or left.
And there are more people leaving the nursing workforce than joining the nursing workforce, which is a reverse of a trend just, you know, come in in the last couple of years.
And lastly on radiologists, how many of the radiologist posts are vacant at the moment? 1 in 12.
So all of those workforce points are really there just to say that it's worth having a look at what more technology can do for our sector and AI offers great promise doesn't it.
Last week's report said that based on current trends the 100,000 gap on the right-hand side could well be 250,000 or 350,000 in ten years' time or so. But all of that is based on the bit in bold on the left-hand side, that the workforce gaps are only going to get worse if current models of care and staffing continue and of course, AI and other technology offers the promise of alternative models of care.
So how much hope can we have in artificial intelligence and augmented intelligence as Daniel mentioned earlier, so support for clinical decision-makers rather than taking over the clinical decision-making? This is the Gartner hype cycle, which for some things that are currently hyped show when they crash and burn and don't turn into anything productive. And on the right-hand side things that make it through to actually making a difference in the real world in that sort of brilliant bit of the curve called the plateau of productivity. All the things on here Gartner say are already making a difference if they're on the right-hand side, or likely to see their way through development and make a difference in future.
This is my version of Charles's slide, machine learning and IoT platforms, much hyped at the moment, Gartner say, in five to ten years' time they will have made it through to the plateau of productivity in the way that the speech recognition has, or is about to make it through. And the chatbots and sort of the new or growing sort of industry around AI governance, which is just good to note from earlier on today and of course, much needed to make current regulation fit for the future.
I thought what I'd do is just highlight a couple of examples of work that's going on in the UK as well as the US that shows that, shows there's plenty of hype about AI in health and care, but there's a bit of substance behind that hype as well and with your support I hope, there'll be more and more to trumpet for the public in this country and for export potential overseas, and whatever else Hassan was saying.
Two examples, both from the report that I helped to compile with some of the people in the audience of the state of play of AI in health and care at the moment. Some of the challenges that those we surveyed said needed to be overcome, or where further work was required, which is sort of Adam's territory as we move on. Two examples, the second one is a chatbot and first one is more predictive analytics.
You might remember from the NHS Five-Year Forward View there was an announcement about NHS testbeds. One of the first seven testbeds was this project, the Surrey Dementia Testbed based in the 5G Research and Innovation Centre at the University of Surrey, to use the internet of things and a variety of sensors in the homes of 400 citizens around Surrey to support people living with mild to moderate dementia and the carers of those living with them to see what the internet of things and AI technology could do to improve their daily lives.
Data comes 24/7 from a variety of those sensors and monitors and wearable devices into a server in the university, they collate the data, cleanse it and provide it out to a clinical monitoring team, with that integrated view on the left-hand side. The predictive analytics looks at what a normal pattern of data is for an individual and highlights anomalies when they come through. A clinician decides what to do with that. And the sorts of things that are coming through in the pie chart on the right-hand side are showing that with the big data analysis and machine learning that's available from 24/7 delivery of data from 400 homes, it is able to detect early signs of a urinary tract infection, early signs of dizziness that might lead to falls, and lead to earlier intervention from GP or other community based worker to help keep people safe and healthy in their own homes.
That's happening now in the real world, as is Oli the chatbot in Alder Hey Hospital. And when I heard the lead paediatrician talk about Oli the chatbot it was a bit of a revelation for me having thought that my interest in AI and health and care was really about helping the sector cope with the workforce challenges that we've currently got and are going to have more and more so in the future. But Ian was talking about how the chatbot was able to improve the care that he had been able to provide face to face, because the children and families asking questions of the chatbot were asking different questions to him than they were prepared to ask directly. So his famous example of that was saying in all his years of clinical practice he'd never had a child say to him what one of the children had asked the chatbot about: "will I wake up from my operation?"
And I think, new technology can improve clinical care so long as we get the public trust and the right use of technology in the way that other speakers have said.