Milan Tjioe (dermatologist) on innovation, resilience and the future of healthcare

Milan Tjioe (dermatologist) on innovation, resilience and the future of healthcare

Dermatologist at Bravis Hospital, co-founder of Webcamconsult, Doctor2go and Derm2check

Milan Tjioe is a dermatologist at Bravis Hospital and one of the pioneers in the field of digital healthcare in the Netherlands. As early as the late 1990s, he was experimenting with handheld computers and digital reference information for doctors. He is now one of the founding figures behind Webcamconsult, Doctor2go and Derm2check – tools that lower the barrier between patients and specialists. In this interview, he talks about the journey he has taken, the obstacles he has faced, and his vision for the future of AI in healthcare.

From illegible handwriting to digital healthcare

For Milan, his fascination with digital healthcare actually began in a very practical way: doctors with illegible handwriting: “It started with me feeling that things needed to go digital. That was long before the pandemic, when digital healthcare suddenly took off. I’d already been fascinated by computers and the idea that you could do so much more with handheld computers. That had already started back in ’98, ’99.”

In those early years, the possibilities were limited, but the ambition was great. Milan and a colleague began actively promoting the use of PDAs amongst doctors, not as a gadget, but as a practical tool. “You could simply carry a huge amount of reference information with you. If you wanted to look something up, you could just do it on the spot, rather than having to search for a book. We spent years promoting that.”

One concrete result was the digital release of the Farmacotherapeutisch Kompas, a standard reference work on medicines that had previously only been available on CD-ROM and in book form. “We thought: surely you should be able to do that on your phone too? Eventually, we approached the publisher and said: look, you can do this as well. And now it’s simply available on everyone’s phone. We were the driving force behind that.”

The origins of Webcamconsult

When Milan started working in Bergen op Zoom, the dermatology department at Lievensberg Hospital consistently ranked highly in national rankings. This attracted patients from far afield, but also meant unnecessary travel for simple consultations about test results: “We thought it was a bit of a shame that patients had to travel all the way from so far away just to get their test results or treatment advice. It was possible over the phone, but you do miss out on something. It’s less personal. We also wanted to get a better idea of how the patient was actually feeling.”

However, the tools available at the time were unsuitable: conferencing software with shared access codes, where privacy was barely guaranteed.

"Imagine this: you’ve got two or three patients one after the other, and they’re all given the same code. In theory, that means they could all be in your digital consultation room at the same time. So that really doesn’t work. The solution: a system with a unique, one-off link for each consultation, without patients or doctors having to install any extra software. That was the idea: a link, click, done. And you can’t use that link again afterwards. Simple and secure.”

The Netherlands is lagging behind, and there’s a reason for that

When asked how the Netherlands is faring in the field of healthcare innovation, Milan is clear: 

“The Netherlands is lagging far behind, for the simple reason that, back in 2014, there was a lack of trust in the national electronic patient record (EPD), which was subsequently rejected by the Senate. That has held back digital innovation. The result: every hospital and every GP works with their own system, and those systems don’t communicate with one another.

Patients think that when they visit the doctor, the doctor knows everything. But we know nothing, apart from what’s in our own system. The rest is restricted. And that makes healthcare more expensive, because we end up taking blood samples again, even though that was already done elsewhere last week. Countries such as the Baltic states and even the Veneto region in Italy solved this problem years ago. By 2008 or 2009, they already had a national or regional electronic patient record (EPR), where all patient information was always accessible and properly secured. Even in Italy, you could collect your medication from any pharmacy in the region, and the details were accessible to all those involved. And we still can’t do that in the Netherlands.”

According to Milan, the frequently cited concerns about privacy are greatly exaggerated. “People have been made incredibly fearful. There have been no cases in the news where the system has been deliberately misused. Of course there are data breaches, but those happen anyway. There are ways to ensure proper security. It’s more a matter of people’s fears being exploited than a real risk.”

Digitalisation: a step backwards after the pandemic

During the pandemic, the use of digital healthcare soared, but this proved to be temporary. “Once the pandemic was over, people simply reverted to their old habits. And then there’s the debate that was recently in the news: GPs are being forced to use digital healthcare tools through negative financial incentives. If you don’t use a particular triage programme, you face a financial penalty of thousands of euros. It’s true.

The result is that vulnerable groups are at risk of being left behind. Coercion doesn’t work. It’s simply being imposed on doctors and patients, and good luck with that. Whereas you really need to guide people through it step by step. We have a ‘Digituin’ at the hospital, with volunteers who guide patients: how do I log in with DigiD, how do I do this, how do I do that? So that eventually they can do it themselves. That’s very important."

The biggest misconception about AI in healthcare

AI is currently the buzzword in the healthcare world. “The biggest misconception about AI is that it’s intelligent. It absolutely isn’t. At the moment, there’s very little intelligence involved. It’s probability: which word follows which word. That leads to a very real risk: hallucinations. Systems make things up because they generate the most probable follow-up sentence based on probability, even if it is factually incorrect.

I saw an example at a refresher course for dermatologists. The file stated: the patient suffered from depression in 1995. What does the summary say? The patient is feeling low and has suicidal tendencies. The system doesn’t understand that this was in the past. The patient was depressed twenty years ago. Not anymore. But the system turns this into a current problem.”

Derm2check: AI that does work, provided it’s used for specific purposes

Milan does believe in AI, but only as a strictly defined tool. That is precisely the philosophy behind Derm2check. “The problem is that people use AI for everything. You have to deploy AI very specifically for a particular purpose, trained on data that relates specifically to that purpose. Then it can really offer benefits.”

Derm2check builds on a decision tree that has been in use at the dermatology clinic for twenty years. Doctors enter symptoms, photos, age and gender and receive a list of possible diagnoses, links to literature and treatment options.

“Important: the system does not make a diagnosis. It provides advice. We tested ten different AI systems to see which was the most reliable, trained on our own data. It produces five suggestions with possible diagnoses and references, so that the doctor can compare them and say: ‘I think that’s the most appropriate one.’”

The need is urgent: there are currently around thirty to forty unfilled vacancies for dermatologists. “If GPs can carry out some dermatological treatments themselves without referring patients, patients will be helped more quickly; GPs will also learn from this, and we as dermatologists will face less pressure from patients whom the GP could actually treat themselves.” 

Doctors remain essential, but they do need to keep learning

When asked about the future, Milan is cautiously optimistic, but also critical of the way AI is currently being used. “Letting AI make decisions simply isn’t right yet. It’s unbelievable that companies are replacing staff with AI. In many cases, things still go wrong. The doctor remains essential.”

At the same time, he sees clear added value when AI is used as a sounding board.

“That combination – humans and AI working together to provide advice – creates synergy. AI might say: ‘This area looks suspicious,’ or ‘This medication might be more effective based on this data.’ And then, as a doctor, you assess: ‘Is that really the case for my patient?’”

"Doctors need to be better trained in what AI is, what it can and cannot do. What are its limitations? Once you know that, you can make good use of it."