Transforming healthcare with AI

As a hub for both Artificial intelligence (AI) and life sciences, Amsterdam is in an excellent position to take great strides in bringing the two together, and medical and data professionals in the region are joining forces to transform AI in everyday healthcare.

The early impact of AI on healthcare in Amsterdam

Artificial intelligence (AI) and machine learning hold great promises for medical care. The combination of data with medical expertise and machine learning can significantly simplify, accelerate and improve diagnoses. The effectiveness of new therapies can be monitored much better. AI can also play an important role in prevention.

Of course, by its very nature, AI as a technology can be applied in practically any field. “AI is not restricted to any domain,” says Cees Snoek, professor of Intelligent Sensory Information Systems at the University of Amsterdam’s Faculty of Science, as well as head of the QUVA Lab (a joint research lab set up by Qualcomm and the UvA for research on deep learning and computer vision) and co-founder of the Innovation Center for Artificial Intelligence (ICAI). But, he argues, “especially in healthcare I see a lot of potential, and there are not too many places in the world where this combination of AI and healthcare can be exploited.”

Jeroen van Duffelen, Aidence

And one field of healthcare where AI is already making rapid gains is medical imaging and analysis. The Netherlands has been one of the world’s biggest producers of AI medical imaging processing technology for decades – there are more medical imaging companies here than in the United States, China or anywhere else in the EU.

“When it comes to medical imaging,” says Jeroen van Duffelen, co-founder and COO of MedTech frontrunner Aidence, “we are a big country on a very small surface.” And at ICAI, Snoek and co-founder Marcel Worring are heading up the AIM Lab, where AIM stands for AI for Medical Imaging. The lab, a collaborative initiative of the Inception Institute of Artificial Intelligence from the United Arab Emirates and the University of Amsterdam, focuses on using artificial intelligence for medical image recognition.

The first wave of startups and academic collaborations

Amsterdam Skills Centre

Amsterdam Skills Centre 

In Amsterdam’s business and ­­­startup ecosystems, AI and healthcare are certainly coming together. “The Netherlands is the new centre of healthcare in Europe, both in terms of data science and quality healthcare and healthcare systems,” says Aidence’s Van Duffelen. His company is developing AI-driven healthcare solutions to improve the early detection of lung cancer.

Then there’s BrainCreators, an AI company that has been working with physicians on a method of analysing brain scans to improve accuracy in diagnosing strokes. By using a deep neural network to identify stroke locations in 3D volume, it has been able to trim the time of diagnosis from six minutes to less than 30 seconds, while also improving accuracy. BrainCreators is also training an AI model to predict the efficacy of cancer treatments.

Another company, Pacmed, is exploring how to use AI to guide doctors to provide the best possible treatment. After more than 100 GPs used Pacmed’s system during an implementation study, the startup is now developing algorithms and software to help treat diabetes, hypertension and kidney failure.

And Kepler Vision, one of the winners of the inaugural Academic Startup Competition, develops body language recognition software that uses AI to analyse video footage. Its star product, the Kepler Man Down Detector, can detect if someone is lying on the floor or needs help and can be used in, for example, care homes.

Research centres in Amsterdam, such as the Netherlands Cancer Institute (NKI) and the Amsterdam UMC Imaging Center, are at the forefront of AI applications in healthcare, too. And the Amsterdam Skills Centre, a public-private partnership of Amsterdam UMC and leading medical technology company Stryker, is a ground-breaking training centre using AI and virtual reality (VR) to train the next generation of surgeons.

Where is AI in ‘real life’?

Marcel Worring, ICAI

One key question remains, however: where can you come across all these ground-breaking innovations today, if you aren’t a researcher, don’t work at a MedTech startup or are a surgeon in training?

It is undeniably early days for AI in healthcare. Many of the scientific advances in diagnosis and therapy are far from reaching daily medical practice. And experts know better than to throw around daring prognoses – “humans are notoriously bad at predicting how fast or how slow technology changes,” says Snoek.

Nevertheless, medical AI applications have been the subject of much conjecture. “There were a lot of big promises from big corporations and small startups about what AI could deliver, from curing cancer to making radiologists irrelevant within five years,” says Van Duffelen.

The hype has settled down but it can certainly be said that in 2019, the capabilities of AI and machine learning technology and applications are light years ahead of what was possible just a few years ago. “Nowadays you really see new things you can do that you couldn’t have done a decade ago. Now we not only understand what is happening but can apply it too, and learn from that. That’s very exciting,” says ICAI’s Worring.

In fact, things are moving along steadily. Right now, Aidence is being used at medical facilities in the UK, Sweden and the Netherlands, and is certified for clinical practice throughout the EU. Amsterdam UMC and Pacmed have signed a long-term cooperation agreement to help further applications of medical data science in a clinical setting. The 11-year deal aims to improve patient care at Amsterdam UMC’s Intensive Care Unit (ICU) by using machine learning algorithms and AI software to support medical professionals who treat critically-ill patients. And Kepler Vision, too, have signed a contract with a provider of care homes.

Building a healthier future with AI-driven healthcare

Netherlands Cancer Institute (NKI)

The hopes of what AI can do for patients and medical professionals remain high, if not far-flung. “I cannot predict the future,” says Snoek. “But I hope that AI will complement the doctors and the caregivers and will help faster turnarounds of data processing. Many doctors have been converted into people who look at screens the entire day and it is my hope to bring them back to the patient. If we can come up with algorithms that release them so that they can get back to their real job, that would be great progress.”

Van Duffelen adds: “Now we are seeing that AI can improve healthcare – making it better, faster, cheaper – but can’t replace medical professionals. We’ve moved past the startup hype to a stage where companies are developing technology that benefits society.”

Snoek mentions using AI to recognise tumour growth or Alzheimer disease in images. “Initially these have very tiny effects, hardly visible to the human eye. And the hope is that these minute changes in images and in videos – because more and more data will be video – can be recognised, that we can point the caregivers or the doctor to it. By doing so, we could signal earlier that something is wrong and provide the appropriate care. That would be my dream.”

Emile Voest, the medical director at the NKI, which has been at the forefront of cancer research and treatment in the Netherlands for more than 100 years, agrees that AI will play a significant role in cancer research. “Images of tumours in response to treatment, sections of tumours under the microscope, blood samples with biomarkers... these are all areas where AI can play an important role. And if you have a database that is really big and combine that with machine learning... well, it’s early days, but I'm convinced that this can make a significant impact in the coming years. Combine all the information we have in data sets, and we can take huge steps forward.”

Dealing with the opportunities and the challenges presented by data

Amsterdam UMC Imaging Center

Amsterdam UMC Imaging Center

Data, of course, is not just the backbone of AI innovation, but also one of the biggest challenges in the field, certainly when it comes to medical data and privacy concerns. “All the recent progress in AI is based on data,” says Snoek. “It is the availability of huge amounts of data that really fuels AI algorithms; that is the guarantee for success. This is a challenge in the medical field, where data is limited and nobody wants to share it.”

Voest, however, doesn’t agree with that last point. Of course, he says, the use of patient data must happen on a voluntary basis, and “privacy is an important topic. But I’ll be honest, if we have a patient and we ask if we can use their data for research, almost no-one says no. All patients want to help their fellow citizens.”

And Snoek concedes that finding a solution is not an impossible task. “I think if there is one place in the world where we can arrange this in the right way, it’s the Netherlands, it’s Amsterdam,” he says. “I think we have a good history of collaborating and really talking to each other – and making sure that things are done in a fair and transparent way.”

Jeroen Maas, Health Challenge Lead at the Amsterdam Economic Board, also notes the region’s strong track record in data collection, research and ethics. “It’s about truly thinking through an entire process, anticipating any issues and questions. When that is done correctly, you end up with intelligent, applicable and actionable outcomes. Ultimately, the use of data is going to strengthen the implementation of fact-based policy making, which is to the advantage of all of the Amsterdam Area’s citizens.”

Work is well underway in the region to overcome the challenge, collecting more and more medical data ensuring it is never misused. For example, in cooperation with the European Society of Intensive Care Medicine, Amsterdam UMC is providing guidance for European intensive care units to share their data in a responsible way. Amsterdam UMC have already anonymised their own database and plan to release it to interested ICUs. And there are talks between hospitals in the Amsterdam region about how to share data amongst them.

How to foster new talent while preventing a brain drain

Amsterdam University College

Amsterdam University College at the Science Park

Amsterdam is working hard to secure a solid labour pool for AI talent. To help nurture the next generation of AI experts, Amsterdam School of Data Science – a collaboration between four of the city’s research and applied sciences universities – offers more than 250 data-science driven programmes. And in 2017, the Growth Tribe academy, which specialises in teaching digital growth skills, launched a new AI and machine learning training programme for people without a technical background – the first of its kind in the world. The city’s two universities also offer a joint master’s programme in AI.

Still, Snoek warns that talent can present another challenge. “We need people who are capable of inventing new fundamental AI algorithms, while at the same time being capable of applying it in the medical context.”

In addition, there is a worry of losing too many people who could work in research or other academic positions to big tech companies. Founding ICAI, he says, was one way to prevent this. “To give talent an opportunity to get the best of both worlds, so they can collaborate with industry.” ICAI ensures academic freedom by protecting researchers' rights to publish their research, while also guaranteeing that the private organisations involved have a first option to buy any IP that might be created through the research.

As for Amsterdam’s future as a leader in AI in healthcare, Snoek is optimistic: “We have the mindset, we have the vibe, we have the skill sets. We have excellent medical skills, we have excellent AI skills, and we have the data that’s stored in all the hospitals; we have the tech and we have the talent. All the ingredients are there. Now we have to make the soup.”