‘We have a chance to affect healthcare on a global scale’: Ross Upton, CEO Ultromics

The number one cause of death for both men and women worldwide is coronary artery disease. It accounts for one in three deaths in adults over 35 years of age. Seventeen million people worldwide die every year from cardiovascular disease. According to the British Heart Foundation, there were 86,474 heart failure admissions in England in 2018-19, up from 65,025 in 2014-15. While the causes of heart disease – none of which are helped by the proliferation of obesity, diabetes, high cholesterol and smoking – are well known, diagnosis using echocardiogram tests has significant room for improvement.

According to Ross Upton, CEO of Oxford-based start-up Ultromics, although figures vary from one source to another, current methods used to assess echocardiograms (heart scans) have a ballpark diagnostic accuracy of around 80 per cent. Upton says Ultromics is working to reduce this diagnostic inaccuracy rate with its AI-based, ultrasound diagnostic software platform EchoGo, which will aid clinicians to increase diagnostic accuracy to above 90 per cent. “Tests need to be more quantifiable,” says Upton. “Then we have a chance to affect healthcare on a global scale.”

While working on his PhD at Oxford, Upton set about developing “a piece of technology that diagnoses coronary heart disease by using artificial intelligence to a greater standard than in current clinical practice”. Today’s main imaging technology is the echocardiogram test that provides views of heart structures. “When you get chest pain – usually it’s nondescript – the doctor will refer you for a functional imaging test, and in the majority of cases this will be an echocardiogram.” Upton goes on to say that the echocardiogram accounts for at least as many tests as “all of the other available imaging modalities – X-ray, CT, MRI – put together.”

“Echocardiogram is a fantastic imaging test,” says Upton. “The whole thing comprises taking ultrasound videos of the heart that get transferred to the doctor’s computer, where the images are analysed in order to produce a diagnostic report.” One reason for the test being "amazing" is that the machinery has evolved “to the point where it can be wheeled up to the patient’s bedside – some echocardiograms are available on mobile phones now. It’s getting to the stage where they are now so portable, cheap and accessible that they are found everywhere. It’s the most popular modality used because it also gives you real-time information on how the heart’s performing. You can’t, for instance, wheel an MRI up to a patient’s bedside, and they’re not available in remote areas of developing countries.”

The problem with ultrasound is “there is lots of variability in how doctors interpret these images. You can make measurements that will give you some indication of a diagnosis. But you can also visually interpret the images. Usually, you do both to come up with an assessment.” Yet there is an element of subjectivity in the assessment that depends on the level of clinician experience, “which leads to a huge variability in diagnosis. The diagnosis can range from 60 to 85 per cent accuracy.” Accuracy is assessed by following up with patients after diagnosis, “and there are two things that can go wrong”. First is a false positive, which results in the patient going for unnecessary and expensive surgery. Second, a false negative can result in a patient requiring surgery being sent home untreated, with the ensuing risk of myocardial infarction (or ‘heart attack’). “When that happens, it can be fatal. If it’s not, it involves sending out ambulances, on-call doctors and so on. That can represent a huge cost to the health system.”

Upton says the most recent available accuracy figures puts the situation at “about 76 per cent (with a huge variation)”. This means there is plenty of room for improvement “and that was the problem that I basically tried to solve in my PhD, long before there had been any intention of spinning out Ultromics as a business”. The title of Upton’s thesis is so long, technical and complex that the author himself cannot be certain he’s remembered it entirely correctly, which is presumably one of the reasons he opted for the much simpler descriptor for the company he founded.

The name Ultromics is a portmanteau word starting with a contraction of ‘ultrasound’ (the ‘ultra’ prefix coming from the Latin for ‘beyond’). “Ultrasound is a massive technology with something like 200 million scans done globally. We didn’t think we’d ever step outside that. We were comfortable being that specific.” The second part of the name – the Greek-derived ‘-omics’ suffix – is a scientific neologism generally attached to emerging ‘large data’ scientific fields, such as genomics.

Upton has spent nearly all his adult life in academia. The 30-year-old British scientist read for his undergraduate degree in zoo biology at Nottingham Trent University. His first Master’s degree is in clinical biochemistry from the University of Surrey, with his second (in cardiovascular imaging) from Newcastle University. This was followed by a PhD in cardiovascular medicine at the University of Oxford, which is where Ultromics started. He cheerfully admits he’s “not been out of the academic world long”. In fact, although Ultromics “has been going for about two-and-a-half years, I only handed in my thesis a couple of weeks ago. Technically, I’ve probably only just left the academic world in the last couple of weeks.” If all goes to plan, he’ll be seeing in the New Year as ‘Dr Upton’.

This path could have been different. By his own admission Upton was “not very good at school. I was wasting a lot of my potential and enjoyed going out partying more than my education. I did terribly in my A-levels and set the record for the lowest-ever attendance in my sixth-form college. I drove my teachers and my parents crazy. I was never bad – I was just having far too much of a good time.” As things do, this all came to "a bit of a head" in the form of a car accident, followed by a family intervention, after which “I started to knuckle down. That’s what led me down this path, and I thought the only way I was going to stick at something was if it was a subject I was passionate about. The only thing I could put my finger on was zoo biology. I love animals and I love conservation.”

During that degree, in which there were modules in anatomy and physiology, there was a lecture on the cardiovascular system, “and that was it. I got a good degree and that was a massive transformation for me, and it led me to want to pursue this further.” While doing his first Master’s degree, Upton attended a lecture by a cardiologist who was also working on a start-up called Cardiocity that had “developed the first touch-ECG (electrocardiogram) using sensors from a McLaren F1 car that they put into an iPad, from which you could get an EGC signal. I was absolutely fascinated and went to work on that project. That was the thing that got me into the cardiovascular and cardiology space at the forefront of innovation.”

During his second Master’s, Upton met his PhD supervisor, ultrasound specialist and consultant cardiologist Professor Paul Leeson, who was to co-find Ultromics with Upton. “During this time, I decided to specialise in echocardiograms because I thought this was the most useful and practical cardiovascular diagnostic technology – I didn’t think anything else came close. I also thought it had the ability to affect people on a global scale. Ultrasound is general and accessible – not a privileged technology.” It turns out Leeson had been collecting a database of ultrasound images "for years" in order to address the problem of variable diagnosis in echocardiography. “Everything came together for me at that point,” says Upton, who goes on to describe the train of thought that if you believe this is the technology, it follows that it needs to be less subjective and more quantifiable.

If the holy trinity of echocardiography is to process images, automatically make measurements and produce a diagnosis, then the holy grail is to produce a single piece of technology that can handle all three tasks. “That’s what we set out to do,” at a time when technological commercialisation of the University of Oxford was booming, with the creation of Oxford Sciences Innovation (OSI) that includes in its mission statement a “commitment to build thriving science and technology companies from Oxford”. Upton says OSI was actively looking to spin-out companies, “and they said our technology looked like it had commercial potential”.

At first, Upton wasn’t interested, preferring to remain in academia and publish on the subject. Yet OSI had the argument that the quickest route to having a positive impact on patients was to commercialise the technology as a business, “otherwise it will just sit in a publication waiting for someone else to do it.” Cautiously, Upton agreed on the condition that “improving patient outcomes came first”. Six months later in May 2017, Ultromics was formed with the academic Upton metamorphosing into an entrepreneur. “Since I’ve been doing this, I’ve never wanted to do anything else.”

When asked if he knew of existing markets, Upton remarks that because of the nature of his research: “I was the market. When I was doing my second Master’s I was using existing software analysis programmes, and they were so bad.” He goes on to explain how state-of-the-art technology was frustrating his research. “It was meeting Paul, and understanding that diagnosis could be improved that led me to suspect there was a huge market.

“EchoGo does all the things I’ve mentioned. It’s a cloud-based software suite that takes ultrasound images, processes them, creates automated measurements and then provides a diagnosis.” It’s a bit like importing MP3s into iTunes or digital photos into post-processing software, says Upton: “The doctor or clinician doesn’t have to do anything. It provides automatic analysis of an echocardiogram and supports diagnosis. It will be targeted at hospitals, and if a hospital wants EchoGo, we just have to switch on a port. They’ll tell us which clinics they want it in, we pull in the images, process them and send back the report.”

Upton recalls “the day we spun-out the company it was just me sitting in an office in my investor’s headquarters. On my very first day we decided we needed regulatory advice for the product we were trying to build. My first job on my first day was to find someone that knew the regulatory landscape.” Having contacted a consultant willing to help, Upton was told he’d be sent a purchase order. Realising he had "no idea" what a purchase order was, or what he was supposed to do with it, the scale of what he had taken on started to dawn on him. “The other thing that struck at the time was seeing the company’s bank balance. I had gone from being a student on a stipend to suddenly having access to a business account” with more money than he had ever seen before displayed on his mobile phone.

After recovering from the culture shock of moving away from academia and into the commercial world, Upton adapted quickly, growing Ultromics originally to a team of seven by the end of the first year of trading. By 2018, following a series of what Upton describes as "good clinical trials", there was a second wave of fund-raising that drew in £11m, that in turn “allowed the business to grow quite quickly. Today, we’re 40 people.” Upton adds there are still strong links back into academia, with the University of Oxford a shareholder in Ultromics. “We also work with them and partner with them for our clinical trials in the UK, where they do most of the work in terms of keeping patient information separate from the company. We in turn fund university personnel to work on that data, and if there’s any IP generated from that data, it will then get licensed to the company. It’s a really good practical arrangement that a lot of companies could learn from, in that we don’t have access to patient data, but we still benefit from research going on in the NHS.”

Ultromics has just become one of the first companies to have 510(k) FDA clearance for an AI medical device. EchoGo will now be available for clinicians in the UK and US to use to help them identify cardiovascular disease earlier. It automates cardiac measurements and is the first AI application to measure cardiac strain. There are approximately 50 million heart scans conducted a year in the US, and cardiovascular disease is the number one global killer.