For our latest interview we spoke to Dr Janak Gunatilleke to find out more about his experiences and particular interests in the world of digital healthcare.
Here’s what Janak had to say…
Can you tell us a bit about your background and your role?
I started off my working life as a doctor, I worked as a junior doctor in the NHS. Overall I’ve got about 16 years of experience in healthcare – during that time I’ve taken on a variety of roles, from consulting roles to operational roles to working for a health tech start-up. I was also the co-founder of an AI start-up based in Sri Lanka called ConscientAI.
Over the last four years, I’ve been focused more on the digital and data side of healthcare. As part of that, I’m currently finishing a Masters degree in data science, and I’ve also been on the evaluation panel for the NHSX AI awards. I’m now a Director at KPMG where I lead on UK healthcare analytics.
What are your main interests in digital healthcare?
I have two main interests, the first is around user-centred design and the second is around how best we can use data and AI to solve clinical problems.
On the user-centred design side of things, I got interested in this and learnt a lot about it whilst I was with the health tech start-up Mindwave Ventures. We were applying these principles in everything we did – whether that was developing a product or delivering bespoke digital health projects for our clients – to make sure we engage the end users in the process, identify the real problems and build better solutions.
Lately I’ve been thinking about how to apply the concepts of user-centred design to clinical pathways. If you think about taking a truly patient-centric view, a patient doesn’t present with a disease like asthma, they present with symptoms such as difficulty breathing or breathlessness. I’m exploring this more in my Masters research, taking a symptom focussed approach to analysing the pathway of patients who present to the Emergency Department with breathlessness. I am looking to understand whether there are any specific patient subgroups, what investigations and treatments they receive, outcomes such as a new diagnosis or death, and whether there are any predictors of the outcomes.
I’m also looking into how we can use process mining, which is a process to get data from electronic patient records, to understand the order and timing of events – like when they had a particular test or when they might have been prescribed a particular medication, to see whether that’s useful. I published the first output of this research recently Colleagues and I did a literature review of all the studies that looked at patients who present with breathlessness and how their clinical pathway was analysed, whether the approach was symptom-focused or taking a more disease-focused approach, looking at whether the patient has a disease like asthma or COPD. We found that the vast majority of studies had a disease-based focus and that a lot of the studies were done in single departments or single hospitals, and there wasn’t much process mining used at all. So, I think we should investigate more the potential benefits of symptom focused clinical pathways and the use of process mining to help understand these pathways by doing larger studies across different healthcare settings.
Coming back to my second interest, how we can use data and AI to solve clinical problems. This is something I’ve been looking at and thinking about over the past couple of years. I think there’s a lot of hype around AI in healthcare and what it can do – but if you look on the ground, there’s not that many AI solutions in widespread use, apart from a few examples in radiology, for example related to diagnoses of strokes.
I wanted to look deeper, beyond the hype to determine where the true potential lies. I also wanted to know the challenges – what’s preventing the good solutions from being adopted and scaled? These are some of the questions that I answer in my book, ‘Artificial Intelligence in Healthcare: Unlocking its potential’.
Can you talk about the projects you are currently working on, and your approach?
In my current role at KPMG, I work mainly in three key areas. The first area is about helping clients think about how to make the best use of their data and to develop an associated data strategy to realise the benefits and improve citizen outcomes. I think it’s going to be increasingly important as Integrated Care Systems (ICS) are set up and they think about their wider corporate strategy, they can take advantage of the fact that they are made up of multiple organisations across health and care – that presents great opportunities to make better use of data. The data strategy work we do help our clients to identify a coherent set of actions that take towards being a more insight driven organisation.
The second area follows on from that, really – once you have the strategy, my role is about helping them implement some of those key actions. That could be anything from helping them get some of the fundamentals right in terms of infrastructure or how they manage data or equipping their workforce with skills and confidence they need.
The third thing is very interesting, it’s around developing our own products. This is to meet some of the common problems we see across our clients, for example we developed a great product that helps hospitals with their strategic workforce planning which helps them understand their demand better, the impact of their workforce retirement rate, and where the gaps are. It helps them understand the costs associated with changes and what they need to do to plug those gaps. Based on that, we’re looking at other areas and other problems where we can develop a product, a repeatable solution where we can help more clients.
In terms of how we approach the projects, we try to take a multi-disciplinary approach. If it’s a data strategy project, we don’t just have data analysts and technologists in the team, we bring in colleagues who have expertise in other areas such as corporate strategy or customer engagement. It’s important to make sure you don’t have a very narrow focus in terms of data or technology, but to take a holistic approach in terms of how to solve problems.
What are some of the key challenges that you have faced?
One of the most common problems I see is when we try to put technology first, this is something I explore in my book. It’s like getting it the wrong way round – thinking that you’ve got a really cool piece of technology and trying to shoehorn it in by looking for a problem to solve. We need to always start with the problem.
I think there are two strands that we need to consider when we think about solving problems – the first one is making sure you have a clear definition of what the problem is, which needs to look objectively into things like how much of a problem it is, when it occurs, in which context it occurs. Rather than guessing what each of those things are, we should be working with domain experts and clients and people who actually operate on the frontline to make sure that we validate that those things are true.
The second thing is that once you’ve worked out what the problem is and where it sits, you need to ask whether it is worth solving. We need to understand what the benefits are of actually solving the problem – does it save time, does it improve patient care, is there a cost benefit?
I think the other thing to consider is that it may be a real problem, it may deliver benefits, but in terms of context, is it an important enough problem to solve? By that I mean, as an example, the NHS is facing a large number of different challenges – where is this problem ranked? Is it going to get the attention and time and resources that it needs to make it work?
If you could fix one problem in digital health, what would it be?
I’ll focus a bit more on the data side. I think one of the key things about getting good solutions and improving some of the solutions we have, in areas like AI, is providing the access to the right datasets for training models, to those who need it.
The ‘data saves lives’ strategy talks about access to researchers and things like trusted research environments and how the NHS can approach it. The other key stakeholders that need access are the innovators and the companies that are developing AI solutions – how can they get access to the data they need, in terms of representative data around the different populations, in an appropriate and safe manner?
Of course, this isn’t easy, there are several challenges that we need to consider. Firstly, you need those datasets that are representative and well-curated, either with real data with appropriate controls or artificially generated data (synthetic data). Another challenge is around appropriate controls and governance. Then there’s the challenge of incentivising the sharing and contribution of data from patients and the healthcare organisations that have the data. That’s why it’s important to demonstrate the value of data and what can be done with it.
It’s not an easy thing to fix, by any means, but I think that even making strides towards fixing it would lead to overall improvements.
If you could give one piece of advice to someone thinking of working in your area, what would it be?
It’s a really exciting area, but I think the advice I would give is related to what I said about problems earlier – don’t get carried away by the technology or what it could do. Always use the problem you’re trying to solve as the guiding principle. Then, only once you understand the problem, assess what technology or solution is the most appropriate.
Sometimes what you find is what a simple solution can actually fix a problem and you don’t need to think about sophisticated options, it might not even need a technology solution to solve it. Think about it in an incremental way, like a roadmap; start with a simple solution that addresses part of the problem, then once you demonstrate that it works and you gain the momentum and the buy-in, then you can think about doing the more complicated, sophisticated stuff later on.
It’s important to bear in mind non-technology factors like stakeholder buy-in, and to consider existing workflows too, clinical and non-clinical. There are established ways of working in terms of how a hospital operates so you need to consider how your proposed solution might fit into that and what changes might need to happen, because if you don’t then you’ll definitely run into problems.
You’ll need support to help with things like resources, change management and deployment, so consider what is required and be prepared to invest in that – don’t just invest in the tech.
Thank you to Janak for giving his time and thoughts; you can read Janak’s Masters paper on analysing symptom-based clinical pathways here and find out more about his book ‘Artificial Intelligence in Healthcare: Unlocking its potential’ here.
