The hype cycle for AI is unrelenting. It sometimes seems as though every second company is claiming its product is an “artificial intelligence-driven solution” for a problem that nobody has. In reality, most people’s experience of AI is limited to ChatGPT, or similar large language models, churning out plagiarised student essays, or fake pictures of celebrities with extra fingers.
For some, however, the hype is real. Deciphex, a medical technology firm based in Dublin, specialises in AI-assisted diagnostics, dealing with the real problem of identifying illness. It has just closed a series C funding round that raised €31 million to finance its next phase of development after six years of growth.
Donal O’Shea, chief executive and a veteran of Ireland’s burgeoning biotech start-up scene, pulled together an A-list of backers for the fundraising, including Molten Ventures, a listed venture capital firm, ACT Venture Capital, Seroba and Charles River Laboratories, a long-time collaborator.
O’Shea has raised a lot of money to fuel Deciphex’s expansion, including more than €14 million in equity in 2022 and 2023 and €10 million in venture debt funding from Claret Capital in 2023.
All of it is going towards getting its platforms, Diagnexia and Patholytix, adopted by the world’s big health systems. The HSE and NHS are already customers, and Deciphex is breaking into North America, getting a toehold in Canada’s public health system and the massive market in the US.
The company is diagnosing about 150,000 clinical cases worldwide annually, using AI to help pathologists identify diseases more quickly and accurately.
“Effectively, pathologists are in the midst of decision making in all aspects of our care pathway,” O’Shea says, explaining what Deciphex regards as its market. “But one of the challenges we see globally is a major shortage of pathologists. You simply don’t have enough people to go around.”
Deciphex aims to solve at least part of that problem by digitising, scaling and distributing diagnoses to a network of 250 pathologists who can then move through the process more efficiently. Instead of sending tissue samples or slides to a lab, specialists can get digital uploads of images to accelerate diagnosis. AI can then be deployed for pattern recognition, again speeding up the process. The ultimate goal is to help clear the diagnostic backlogs and pathology waiting lists that virtually every health system struggles with. O’Shea says Deciphex increases productivity by up to 40 per cent.
“I would say it’s the Uber for pathology,” O’Shea adds. “You’re dealing with a digital paradigm and you’re focused on a highly ergonomic way of presenting data to the pathologist. That adds a lot of productivity.”
The company is expanding rapidly. Revenue is more than doubling year on year, according to the company’s most recently filed accounts from 2023. Sales for last year should come in at €13 million or more.
The UK’s NHS is a big customer, accounting for nearly half of turnover in 2023, but O’Shea says the US is the company’s most important market. Headcount, and the wage bill, are growing in line with sales, with employee numbers almost doubling from 2023 to last year.
One might think O’Shea was white-knuckling the rise but Deciphex is not his first rodeo. Although he trained as an academic, he has worked commercially in the digital pathology arena for more than 20 years. O’Shea founded his first company, Slidepath, in 2003, when artificial intelligence was a concept more fit for science fiction than venture capital.
Donal O’Shea became an entrepreneur almost by chance BRYAN MEADE FOR THE SUNDAY TIMES
Slidepath was a campus company at Dublin City University (DCU), where O’Shea had taught since 1999. With a couple of biotech postdocs under his belt, he was on track for a fruitful career in academia. Becoming an entrepreneur was something of a whim.
“I suppose, along the way, I had graduate students, excellent students I worked with, developing them, bringing them through, graduating them with their own PhDs,” he recalls. “One of those guys ended up looking for something to do and we thought, in our wisdom, that it would be a fantastic idea to set up a campus company.”
O’Shea didn’t quit the day job right away. He continued with his responsibilities at DCU while building up Slidepath, eventually selling it to Genetix, a UK company, in 2009, just after the economic crash. It was a good time to exit, as start-ups were finding it almost impossible to recapitalise at that point. But the call of commerce proved irresistible. O’Shea joined the Genetix management team that in 2010 sold the company to Danaher, a multinational based in America, in a deal worth $102 million. Leaving academia was part of the conditions of the Slidepath sale to Genetix. Having proved that he could start, grow and sell a business, though, it was the right time, but it wasn’t easy.
“It was like leaving the civil service,” he says. “It did feel like a big leap at the time. As much as I loved the academic phase of my career, I think what happened subsequently, in terms of going from a small campus company to working with one of the biggest multinationals in the diagnostics arena, was just life-changing and ultimately set me on a completely different trajectory.”
O’Shea stayed in the Danaher orbit for years, with several senior roles in its subsidiary Leica Biosystems, before founding a clutch of start-ups. In early 2016, he launched two, Accelesys and Oncomark, which was sold to a Danaher company in 2021.
Deciphex was founded in 2017. To begin with, Deciphex consisted of just O’Shea and his co-founder, Mark Gregson. It now has 200 employees and continues to grow at a breakneck pace. In addition to the direct employees, Deciphex has a network of pathologists from around the world, effectively making it a very large pathology outsourcing company.
“It’s starting to feel and behave like a scaling, global business,” O’Shea says. “One of the great things about being an Irish company is you’re always seeking critical mass overseas, which makes you drive to be an international company out of the gate.”
At home, Deciphex is taking on extra clinical pathology capacity for the HSE, helping it manage diagnostics and reporting to alleviate bottlenecks in the service. Abroad, Deciphex is pursuing what O’Shea calls “parallel market entry strategies” in places such as the US, Canada and Japan.
“We’re doing a lot of things concurrently, rather than sequentially, like you might if you had less capital available,” O’Shea says. “We still feel that for us to put the afterburners on, that we need to continue to lean into that external capital market to drive growth and ambition.”
He is clearly an enthusiast for the technology and its applications, as he must be, but the use of digital technology in healthcare raises ethical questions regarding equality of access and care. It’s easy to imagine a model developing where one set of patients gets access to medical expertise while another gets sent to Dr Chatbot and Nurse Algorithm.
“Look at things like autopilot,” O’Shea says. “You still have pilots, they’re still flying the aircraft, but the aircraft are bigger, they carry more people, they run on time. And they have something like one failure per ten million. We want to make diagnostics that safe, but able to travel at scale, to handle more capacity. AI plays a huge role in that.”
What about inaccuracies in generative AI that many people have encountered while experimenting with ChatGPT and the like? Are the mistakes evident in large language models a problem in other forms of AI?
“AI will be held to a higher bar than human beings any day of the week,” O’Shea says. “But human beings get tired. Human beings have good days and bad days. And weird biases. At least with AI, you know you’ll have a consistent level of performance.”
He points out that simply training more people is not enough to resolve the supply-demand mismatch in healthcare, especially in some poorer countries where there may be as few as one pathologist per million of population.
“Our view on this has gone way beyond simply looking at building a good business in the UK and US, we are looking at this as a global problem,” O’Shea says. “How do we actually crack the nut across multiple dimensions, both from the biggest payers in the world all the way down to people who can’t afford access to a routine service?”
The life of Donal O’Shea
Age: 53
Lives: Co Meath
Family: wife and three children: “One adult child, two very soon to be. That’s a nice place to be at the moment”
Education: bachelor of science and a PhD from Dublin City University in biotechnology; postdoctoral research in pathology at University College Dublin
Favourite films: No Country for Old Men and Quentin Tarantino’s films of the 1990s
Favourite book: Steve Jobs by Walter Isaacson
Working day: I really enjoy what I do. That’s the first thing. I wake up every Monday morning with a smile on my face, you know what I mean? My day is largely engaging with the commercial and operational people who support me in running the business, and then engaging with my board, with my investors and with prospective investors to look at future financing or strategy. There’s an element of being always on, but I don’t cram my diary day to day with lots of meetings. I try to leave as much space and time to react to situations as best I can.
Downtime: I have very, very good compartmentalisation skills. Even when I work at home, I can close the office door at whatever time so I’m relaxed and spend time with the family. Steve Jobs was always obsessed with walking. So was Einstein. There’s no problem that can’t be solved with a good walk. We’re fortunate to be beside the beach and so you can walk the dog and reflect on the day and process what needs to be processed. I suppose beyond that, it’s the usual kind of stuff like golf, TV, movies, whatever.
SOURCE: The Times
