Lessons from the world’s leading AI-enabled health systems

What is the UK doing well through international eyes? 

Micky Tripathi brought expertise from his work as Chief AI Implementation Officer at Mayo Clinic, a large integrated health care system in the US, and from his previous role as Chief AI Officer at the US Department of Health and Human Services.

Reflecting on what the UK is getting right, Micky highlighted the quality of data. ‘The US relied on UK data for COVID-19… we based everything on the quality and the volume of the data that was available here in the UK because of the way the system is structured here.’

He also highlighted the alignment inherent in the NHS, without the tension in systems where payer and provider can be at odds, praising, ‘a perspective which is taking into account the patient’s health over their lifetime, and can think about investments from that perspective.’

The idea of the power of one system was echoed by Wonchul Cha, emergency physician and Chief Information Officer at Samsung Medical Centre, Korea. He described the impression he had from the event that the NHS is, ‘like a collective intelligence – everybody thinking about how to grow as a system, not as individuals or individual institutions.’

Davina Lim is the Director, Future Systems & Innovation, at the Ministry of Health in Singapore. She said she was pleased that the discourse around AI in health care in the UK was focusing on, ‘evaluation, unlocking the evidence, and making sure that the evidence can travel’, because sharing evidence is an important part of decision-making and wider communication to health professionals and patients.

Bob Klaber, paediatrician and Director of Strategy, Research & Innovation at Imperial College Healthcare NHS Trust in London added that the UK’s universities are, ‘an extraordinary set of assets’.

What are our shared challenges? 

One of the challenges shared across countries and health systems is evaluation and governance of AI tools. 

For Bob, one of the issues around evaluation is being honest about the baseline comparison. Some AI tools take pressure off overstretched clinicians, where pressures are creating risks that may not be acknowledged. In assessing innovations, there are parallels with existing technologies, but, ‘the way AI moves and learns and changes, that is different’.

Panellists spoke of an outpouring of innovation and creativity from health professionals given the opportunity to work with AI, but, as Micky put it, ‘where we’re faltering is the ability to appropriately vet those solutions as they’re coming through. We haven’t calibrated our governance to be able to match that.’ Mayo Clinic is establishing a process to have a rapid cycle review of applications for AI-enabled solutions, to get them to clinical practice as quickly as possible.

For Davina, this is also a key issue. As a funder of innovation, the Ministry of Health in Singapore is asking itself how to decide what to pilot or scale up. They ask three or four questions of a proposed innovation. How enduring is the impact to the system going to be and for whom? If it starts with a single disease pilot, how will the learning be used and how much is repeatable elsewhere? Is the idea reinventing the wheel – has it been done somewhere else? And what is envisioned as the end state of a particular project?

She says, ‘We must support innovation, and we must then think about how it scales across the value chain.’

Wonchul described how the decision to use or buy AI requires different expertise – how to assess and evaluate it is different. More like buying software, the decision may affect different departments and needs a system of governance that can take a longer view.

Where do we go from here and what are the priorities for action?

There are foundations that need to be in place and which are key for the NHS – digital maturity and infrastructure, but also building a culture of innovation. 

Bob reflected on the impression the other panellists had of the NHS working as one system, calling it a, ‘pep talk to all of us who work in the NHS… we have an amazing opportunity around that. Let’s stop behaving in a fragmented way.’ Bob called on us all to consider the data and opportunities we have to tackle health inequalities and to prioritise user experience. ‘As a staff member working in the NHS, there’s been nowhere near enough focus on the user experience of digital-based systems.’

For Davina, the important thing is to focus on communication. ‘It's about trust – it's trust with our health care professionals, that's first, and then after that, it's trust of the public.’

Micky encouraged everyone to decide where they want to be on the AI spectrum, perhaps just watching and considering what becomes available, ‘or you could be at the Mayo end, which is every single day we’re asking ourselves, where are we worried too much about misuse and not enough about missed use? What’s the opportunity cost to a patient of not getting this solution into our clinicians’ hands?’

This content originally featured in our deep dive newsletter.