The last two years, in particular, have been such a blur that perhaps it’s a good time to take stock and think back on what we’ve learnt since beginning the company.
No 6 St Chad’s Place, champagne and Technology Strategy Board
In December 2011, I met my original co-founder, Julian, at a pub in Kings Cross. We were trying to decide whether we should quit our jobs to start Skin Analytics.
On one side was a strong desire to do something that mattered and a belief that we could really improve skin cancer care – getting better outcomes for patients. The original plan was to build a computational diagnostic service from images and use that to dramatically improve access to skin assessments. That in turn would tackle the delay in finding skin cancer and improve patients’ lives.
On the other side, we had no money to start it and we had good paying jobs that helped to support our families.
We had been fundraising for a few months and had received two very clear messages; there would be no money without committing full time, and we needed an anchor to get our initial commitments to come through.
About an hour into the discussion Julian came back with a bottle of champagne and his phone held out with an email on it. We’d just won a Technology Strategy Board (now Innovate UK) grant which provided the anchor all our seed investors would get behind.
I wish I could bottle that feeling. It was the promise of the path less travelled, the opportunity to create something that mattered and relish the challenge that it would entail. My complete naivety of what we’d set out to do is something I enjoy thinking about now.
Our investors weren’t naïve – they knew all too well the challenges we faced and I can’t thank them enough for wanting to power the Skin Analytics journey.
Too early but is there ever a too early?
It’s often hard to put into context memories – especially now that AI has been through a good chunk of the hype cycle.
In January 2012 though, AI hadn’t come of age yet – certainly no one was using the phrase. To put it in context, IBM’s Watson was the “AI” poster child having just won Jeopardy using a set of if-then rules.
At Skin Analytics our ambitions were miles out in front of the technology but we were lucky enough to have the guidance of the late great Professor Bill Fitzgerald at Cambridge University who encouraged us to find a way.
Back then we used classical machine learning – in effect measuring a bunch of things about skin cancer (like shape, colour etc) and then using machine learning to optimise the weighting for each of the features. It worked well in the lab but not well enough to be useful to clinicians.
The paradox of being a founder and our wilderness years
This led us to my first major mistake. As a founder, there is a challenging balance of listening to the experts to learn and find valuable solutions, and ignoring all the experts who tell you why what you’re doing will never work.
In 2013, we were working with clinical experts who taught us so much about skin cancer and the pathways patients followed. They also believed that skin assessments should be left to clinicians and technology would be better at analysing sequential images over time for change – a major risk factor for skin cancer.
We pivoted and focused on that, built a great working model and secured a patent for how we did it. It should have been a great success but, despite engaging users in a free monitoring service, we couldn’t get people to pay for it! It was, and is, clinically useful but there was no future in it when we released it.
To date, the most successful use for it was me catching my wife, Pip, stealing my Easter egg over several days.
What happens when you don’t stay true to your beliefs
While this was happening, large corporates started actively engaging with start-ups to power their continued growth. The mismatch of working with one was nearly the end of Skin Analytics.
We won a global competition run by Unilever to work with their billion dollar brands. Winning that competition was some of the best work I’ve done with a team in a few short weeks and it’s legacy is felt today. It also involved a lot of us staying in a freezing yacht in Barcelona harbour (there we no hotel rooms) and a lot of celebrating when we won.
We effectively pitched to help monitor dry skin for Unilever and built an AI system which consistently outperformed their clinical graders on the severity of it. We were planning on launching the service to 19 markets all at once and the future looked bright. A small problem was that no one had asked their moisturiser customers if they wanted it.
It seems obvious now that if you have dry skin, you put on a moisturiser and forget about it…
The fallout was that we had hired too many people on the promises of our future partnership which disappeared overnight. For Unilever it was an interesting idea that had great executive buy-in but didn’t work out – these things happen and on with the day job. For Skin Analytics though, it was nearly terminal.
It was also the best thing that happened to me and a lesson to stay true to your passion. No one as Skin Analytics was initially focused on dry skin – there were just bigger problems we knew needed fixing and that we knew we could help achieve.
With that, we went back to the drawing board and thought about what we were truly passionate about – getting better clinical outcomes for skin cancer patients.
We were saved by our investors (you know who you are and thank you again!), who bought us time to get back to our very first idea; identifying skin cancer through analysing images.
By this time, we had started working with deep learning and we made a crucial hire to lead our AI efforts in Jack (our AI Director). Immediately we saw performance improvements which took us over the threshold for clinical utility as a decision support tool.
To build a good data pipeline and show some market adoption, we convinced Vitality Health to partner with us to build a teledermatology service (thank you Keith – a real innovator if I’ve ever met one).
One of our brilliant early employees built and launched the service in 6 weeks, on his own, and to this day included some of our best features!
our first ever case.
That was the catalyst for Skin Analytics 2.0 and since then we’ve enjoyed great success. We had worked out who we are and that allowed us to make some big bets early that have stood the test of time. We invested a very high amount of our funds in building clinical evidence and leaning into regulation rather than avoiding it.
These are now the pillars of our company and are increasingly acknowledged as the basic foundation to operate in the healthcare market. That wasn’t the case when we made those decisions. At the time, we were advised to avoid clinicians and the required clinical evidence, skirt the regulations and launch an app!
We didn’t go down that path, instead we doubled down our efforts and we published the first ever powered prospective clinical study for AI in dermatology. It’s a real credit to our Research Director, Helen, that the protocol that she led the design of in 2016 was so relevant the day we published in JAMA at the end of 2019.
We quickly followed up as COVID-19 hit by launching the world’s first AI skin cancer clinical pathway with our NHS partners. We have now seen more than 16,000 patients with pathways that are 99.98% sensitive for cancerous and precancerous lesions.
That’s only possible because of the amazing people who have rallied to our mission.
People like David, who nominally is our Finance and Operations Director but you’d be hard pressed to find a part of the business he’s not involved with.
Or Yasmine and Saeef who have kept our patient facing services thriving during the pandemic through their pure commitment to delivering a great service for our customers.
Or Jim who has brought calm and purpose to our tech delivery in the face of huge demands and while simultaneously implementing both ISO 13485 and ISO 27001 processes into the tech delivery.
Or Venki with his top class QA and entertaining poetry. Or James with extraordinary ability to define a simple but powerful regulatory strategy in a complex world. Or Theo who spent his first xmas with us doing a data migration to avoid Brexit issues. Or Marcello with his AI creativity and magnificent beard. Or Dan and Jenny for their clinical leadership to continue to steer us true to our mission to put patients first. Or Tim, Sarah, Dilraj, Swetha, Fin, Tom, Catalina, Ellie, Giacomo…
I can honestly say I have never worked with such an incredible group of people. The names above are just a sample – I asked our team to nominate people for awards at the end of last year and every single person in the company was nominated!
Since that painful reconnection to our mission, we’ve found nearly 50 people who are committed to our mission to improve patients’ lives and do it in a way that is sustainable to healthcare systems around the world.