The November edition of the Amsterdam Medical Data Science meetup saw data scientists and medical professionals come together at Amsterdam UMC to share insights, experiences and research in medical data science. The events aim to bridge the gap between health professionals and data scientists by bringing both together in an informal setting for presentations and pizza.
The session’s host, Dr Lucas Fleuren, introduced another full meetup by talking about a recent trip to the European Society of Intensive Care Medicine’s 31st Annual Congress in Paris. “It actually had an entire lounge session dedicated to machine learning in the intensive care unit. That conference is intended to be very clinical, so it’s nice to see that it has a dedicated area featuring talks about data science and machine learning in intensive care medicine,” Dr Fleuren said. “Slowly but steadily we are seeing that these techniques are finding their way into the clinic. The gap between medicine and data science is getting smaller.”
A new model for data science research
Facilitating data science research and development in medical centres was the focus of a presentation by Daoud Sie, an expert on molecular biology, bioinformatics and IT infrastructure who works within Amsterdam UMC’s clinical genetics department. Sie introduced his talk by getting attendees to agree that there is “so much more to gain” by using big data in a medical setting, introducing a quote from the Kennisagenda Personalised Medicine (governmental report) which states: “We need substantial investments to provide a solid data infrastructure, and also provide the means to develop new techniques and methodology.”
Together with his colleague Dr Martijn Steenwijk, Sie secured an innovation grant from Amsterdam UMC to develop a sustainable way to help all researchers carry out data science studies more easily. The plan aims to overcome several challenges that hospitals and medical academic centres face in terms of medical data research – including privacy of data, IT infrastructure and educating policy-makers to understand challenges, solutions and the potential benefit of these studies.
Sie said that by overcoming these challenges, hospitals and research centres can use the huge amounts of data generated in hospitals more efficiently for research purposes. Methods to help do this included the introduction of cloud computing, Sie said, as well as developing a community of special interest groups within a centre, where professionals can share ideas and research.
Creating new connections for knowledge sharing
Speaking after his presentation, Sie said that he hopes that the model he has helped develop could be transferred to other medical centres. “That’s very much the plan. We went to other university academic medical centres to investigate what they did there, and we often saw a divide between research and the hospital,” Sie explained. “Our approach aims to solve that by connecting departments – the IT department and the business intelligence departments, say – and connecting organisations as well. It won’t happen overnight, but we have a road map now to hopefully get there.”
Bringing doctors and data scientists together to prevent problems
The second presentation was by Dr Hine van Os from the neurology department at Leiden University Medical Centre (LUMC). He is currently working on a machine learning model to help provide GPs with a better risk assessment of stroke among women. During the presentation he focused on the epidemiological implications of his study, which is using data taken from more than 2.5 million patients.
Dr Van Os emphasised the importance of always being aware of hidden systemic biases in data when developing a machine learning model. He also stressed the importance of collaborating in a multi-disciplinary team, citing the inclusion of a vascular neurology professor, GPs, data scientists, statisticians and epidemiologists in the project he was involved in. “It’s very important that all these people are together in one room once every couple of months to really look at these challenges from all different perspectives,” Dr Van Os said. At the moment, the model is still being developed, but it’s hoped that within the next two years it can be tested in GP practices.
Speaking after the presentation, Dr Van Os called events like the Amsterdam Medical Data Science meetup extremely important. “The match between doctors, data scientists and supporting scientists like epidemiologists and statisticians is not currently optimal,” he said, “and that can result in machine learning models having flaws, or models getting published with shortcomings.”
Big successes in a short time period
As always, after the talks, attendees shared ideas, stories and connected over pizza and soft drinks. Reflecting on the event’s success, Dr Fleuren said that the recent editions had already helped to produce new research projects and connections between data scientists and medical professionals. “It’s only the fourth edition, but already the Amsterdam Medical Data Science meetups have been a great success, helping data scientists, doctors and medical professionals really get to meet and know each other. The talks are great but when people come together at the pizza afterwards, I think that has the most added value. We’ve already had so many people tell us they met here and have started collaborating, and we’re receiving many emails from people wanting to get involved in new projects.”
The meetup takes place on the third Tuesday of every month at the VU University Medical
About Amsterdam Medical Data Science
The Amsterdam Medical Data Science meetups are supported by The Right Data Right Now consortium, which includes Amsterdam UMC, OLVG, Vrije Universiteit Amsterdam, Pacmed and the Amsterdam Economic Board.
For more reports from previous Medical Data plus Pizza Meet-ups, click here.
Tekst: Alex Hibbert – Edenfrost | Fotografie: Brenda de Vries