Meet Simon Rasalingham, CEO of Behold.ai, one of the most innovative start-ups transforming healthcare using AI in the world. The company was recently picked as a winner in the NHSX and Accelerated Access Collaborative (AAC) AI in Health and Care Award bid. In this article Simon tells us about how AI can support clinicians, the importance of asking the right research questions, and what can be learned from Cristiano Ronaldo.
Tell us about your innovation – what and why?
Effectively what we’re doing is speeding up the diagnosis of lung cancer. A lot of people who have lung cancer are picked when they have an x-ray it’s usually an incidental finding. The patient then moves on to having a CT scan and then a biopsy, and it’s a very lengthy process once there is a suspicion.
What we found is that the actual x ray is one of the key points where cancers are typically missed. And when you look retrospectively 90% of missed diagnosis is happened at the X ray point. Why is that? It’s often a very subtle finding an x ray, it’s often missed or there are other things there which may be distracting the radiologist.
And these are radiologist that have usually trained for over 10 years, and they find it still very difficult after 10 years of training. So there’s a great opportunity here to use technology to help the clinician diagnose that there is a suspected lung cancer here and fast track that.
The basis of the whole technology is around using AI based technologies, which are called neural networks, which are a form of looking at the image just as a human does.
And as the way the human recognises the pattern, an algorithm called a computer vision algorithm is able to look at the pixels, and actually diagnose and recognise the patterns like a human does.
We’ve been able to show that the algorithm is actually very good, in conjunction with the radiologist, for detecting those cancers. We can reduce missed lung cancers from the studies that we’ve done by over 60%.
And also it’s very quick. The X-ray is sent through to our algorithm and we then send the result back in 30 seconds. we can then fast track that patient. Rather than sending them home you can say wait here, let’s book a CT scan and let’s get you on the pathway to determine whether it is truly lung cancer or something else.
So, what we’re trying to do is compress a pathway to finding out where you got lung cancer from up to 24 days all the way down to two hours, and that really gives a great opportunity to track these cancers early.
What was the lightbulb moment?
I’d set up a business previously which was the largest tele radiology group called Medica, which is a great success and is still going in. But I wanted to find a new, interesting challenge.
So I started reading all this research on artificial intelligence coming out of America, and there was a lot of hype about how AI is going to replace radiologists, and I thought ‘I don’t think so’.
But there was some fascinating content in these research papers, and I went to my clinical team and to a chap called Dr. Tom Nauton Morgan, who I’ve worked with for nearly 16 years, and like me he thought it was really interesting.
But then he said that the research was missing the most important finding, which is that if you want to build an algorithm that’s good at detecting abnormality, you’ve got to be good at normal. So I realised that the researchers hadn’t been looking at the right problem, and that they weren’t actually talking to any doctors.
Through my Medica days I knew loads of highly skilled radiologists who I could talk to work out the problem, so I just kept going. We worked on the problem together and then once that lightbulb moment came I said right, I know how to build an algorithm, I know how to do it.
It’s been a fascinating journey. In terms of taking out those normal cases, you know, on a like for like basis we’re 40 times better than the radiologist. So the strategy, the plan, the light bulb moment, was correct.
And because we’re very good at taking normal cases out, those hard abnormal cases like lung cancer sprung up as an opportunity.
What’s been your innovator journey highlight to date?
I think it has to be winning the Health In AI Award. We actually pitched for it the first time round and we didn’t win, and I was pretty upset, and had to build the team back up to say that we’re going to go for it again, but we’re not going to go for the phase three, we’re aiming for the maximum national rollout.
And so, building up the team to actually do it again and get there was a big thing, and then we actually won. To describe the the feeling, do you remember when Cristiano Ronaldo lost the European Cup and was crying. And then when he won the European Cup and he was crying.
And I totally understand how he felt, because losing is terrible but winning is the same sort of emotion. I was really I was very happy with that because it was a real justification of what we did, but also as a UK company you know we have taken a very different approach to typically everyone else in the industry, we’re still privately funded, there’s no venture capital in here, 40% of my investments come from clinicians, so they really backed me to go for it and it was a nice win and showed that actually we are the best in class here.
How has KSS AHSN supported you?
When we lost the AI Award the first time round they were emotionally picking me up, and they were as disappointed as I was, so it was like a shared emotional journey. And then they were as happy as I was when we won. When you see other people feeling it as well you realise you’re all in it together.
And it was their encouragement to go for the award again. We connected up with Wessex AHSN as well and we had the both AHSNs helping us out. And that was really helpful, driving it and getting support and feedback on what would be the optimal way to actually win. And we really took their advice on board.
They understood how difficult it was to deploy technology into the NHS, and my argument is if we can deploy a vaccine to over 60 million people in 12 months, why should technology be this difficult.
It shouldn’t, it should actually be easier, so there is a lot of work that we need to do as a as an organisation, with AHSNs, the ACC and NHS X to actually fast track these technologies, because you can deploy them safely.
What has been your toughest obstacle to date?
When I created my first business, tele radiology, a lot of radiologists were saying ‘we’re going to lose our jobs, you’re taking the work from the hospital’. The reality is that no single radiologist has been made redundant in the UK because of these new technologies.
So it’s about trying to sort find out what people were worried about, from a clinical perspective, and then understanding and making sure that they understood the message that this AI technology actually makes them better doctors. It’s not actually taking anything away, it’s actually making them better than they could be without it.
Once we were able to find that narrative we were able to address a lot of the scepticism and fear.
Hopes for the future
The main thing is about deploying this technology and helping address early detection of lung cancer. We want to get to a state where the autonomous nature of this algorithm produces the report, and when it says that the x-ray is normal, that the patient then trusts the algorithm to be correct.
That will be the golden moment, when actually people accept that the algorithm has said that they’re fine.
If you look at it in the context of the system, that means that potentially over three million x-rays could be looked at by my algorithm, instantaneously, each year. That adds enormous clinical capacity back to the system. With the backlog that the NHS is currently facing because of COVID we need to deploy this technology with speed, and we’re very much looking forward to doing that.
A typical day for you would include
My job is to really ensure that the vision that has been created and defined is enacted, and it’s about being able to make sure that the team is aligned towards that vision and growing this business.
We’re a UK start-up and we want to be a global player in this marketplace. We are leading the field in terms of AI diagnosis – we are the first regulated AI company in the UK. So we’re really pushing the safe deployment this technology, and that requires me now to, to educate the NHS, to educate government that actually there is a technology out there we can deploy.
It’s about showing that there is a serious UK player here with global aspirations that can actually lead the world in this technology. I’m very focused on working with the NHS first, showing that our innovation delivers significant benefits, and then rolling it out globally.
What’s the best part of your job now?
I get to meet really amazing, brilliant, kind, decent people, including all the people at KSS AHSN, that have an active interest in getting this done. They find it fascinating and can see the opportunity that actually delivers significant benefits.
What three bits of advice would you give budding innovators?
Never give up. That’s definitely the first one.
The second would be to use your network. I think people can underestimate the power of their network and who they’ve got around them.
My third piece of advice is that you’ve got to put your money where your mouth is. You have to just go for it, you know, you really have to go for it. All of my executive team have put money on the table, and I think that when you start to build that culture, with people committing finances towards an idea, then you create a team that’s highly cohesive.
Find out more about Behold.ai and follow them on Twitter
To find out about the support available from our industry engagement team, please get in touch via kssahsn.bridgingthegap@nhs.net
