Editorial – Collaborate to innovate: driving the adoption of AI in healthcare
In an interview following the EIT Health Think Tank Round Table Meeting in Ireland, Donal Sexton, Consultant Nephrologist & Adjunct Assistant Professor of Trinity College Dublin, offered his view on how to foster alliances between innovators and healthcare professionals (HCPs) that will push forward AI projects, as well as how AI will support collaboration between healthcare teams.
As AI continues to advance, it will potentially transform the healthcare space particularly, according to Donal Sexton, from an operational perspective. He believes patient-facing AI tools are still a long way off and there are several barriers to overcome first, not least the reliance of some hospitals on paper records. But collaboration between innovators and clinicians, underpinned by leadership from regulators and policymakers, will harness the full potential of AI for HCPs and their patients.
How might AI have an impact on how healthcare systems operate and internal teams collaborate?
I think AI could potentially transform operations and processes within healthcare. For example, at Trinity College Hospital, we’re leading the way with the integration of electronic systems that support efficiencies within the hospital.
For example, if somebody orders a particular clinical task, such as a blood test, at a particular time, we can use AI to track how long it takes and the patterns behind any delays or challenges. We can then use that data to identify opportunities to streamline and develop efficiencies across departments. This has been particularly helpful in spotting delays and troubleshooting, and it’s starting to be adopted by others across Ireland.
Other big organisations, such as the U.S. Mayo Clinic where I used to work, are trying to use similar approaches to manage large workforces. Such organisations are aspiring and attempting to use AI for scheduling, managing absences and sick leave and so on, because that can all be very tricky with a huge workforce.
What other uses do you foresee for AI in healthcare?
I am aware that some of the universities here are moving towards implementing AI to improve teaching of medical students. To do this, they collect data about student preferences, the way they want to learn, the most efficient ways of teaching students and so on.
I think the patient-facing part of AI, is still a way from actually being applied due to several challenges. However there is potential at St James’s Hospital given our electronic patient record system.
What are those challenges to the adoption of AI in healthcare settings?
In terms of AI as a medical intervention for patients, part of the problem is in testing and licensing algorithms. There has been a move in the medical literature towards testing algorithms, as you would a drug, in a randomised fashion with patients. The U.S. Food and Drug Administration (FDA) is looking at creating guidelines for that kind of testing. However, because algorithms learn from the data they analyse and change with time, rather than a ‘typical’ licensing route, regulators may have to look at innovative processes to continually monitor algorithms.
We also need to account for things like the specificity of an algorithm’s training data – an algorithm that was developed and worked well in San Francisco in one hospital using data from a specific demographic might work very poorly in the West of Ireland, where the population demographics are different. For this reason, it’s hard to have a ‘general’ algorithm.
But in my opinion, the biggest barrier to adoption of AI in healthcare is actually the lack of electronic systems. A lot is still paper-based and hospitals such as where I work, which is entirely electronic, are the exception. When you have a paper-based system, it’s much harder to analyse the data. It’s possible, but in my experience very difficult. This can also mean that people in different departments of different hospitals are working in silos. It’s much more difficult to collaborate and small projects, even if successful, fail to scale up.
How can greater collaboration between innovators be achieved?
I think one useful way is to centralise things. Having one person within the Health and Safety Executive (HSE) in Ireland for example, who is identified as the lead person for AI innovation in healthcare would make it easier for companies or small and medium-sized enterprises (SMEs) to find opportunities to collaborate.
Because, as I mentioned, lots of people in different hospitals, in different departments within each hospital, have ideas. And then, if they’re motivated, they forge links themselves. But a lot of this just dissipates because there’s no momentum, it’s such a small-scale project and there’s no funding behind it. So maybe those projects could go through a central system at the HSE, who might be able to identify people with whom to collaborate. This could streamline things a bit and avoid everybody duplicating the same projects at multiple different locations.
I think that’s the first step, and as time goes on the HSE will understand the value of that project and put more resources into it.
What leadership is required from government or the EU to ensure change happens?
One main obstacle is funding. Healthcare budgets need to be set aside to progress paper-based systems to electronic systems. If stakeholders think there is funding specifically for that reason, that they can obtain, they’re much more likely to do it than if they have to decide whether to do it within the budget they already have.
And you need to get people involved who are in the position to change policy. Just like the Round Table Series, going forward we need a board of people like that who are involved nationally in pushing for change. And in my opinion, you have to get national governments involved too. Without their buy in, change won’t happen.
Donal Sexton is continuing to work to nurture the sort of collaborations that he believes will be necessary if AI is to live up to its promise, as well as pushing for change by taking part in activity such as the Think Tank Round Table Meeting in Ireland. Join the conversation and stay up to date with the latest thinking in AI on our Twitter and Facebook channels, @EITHealth, using the hashtag #EITHealthAI.