AI, Amsterdam, Health
AI & Health in Amsterdam
This is surprising, given that the amount of data collected in the medical domain has been, and still is, increasing at a rapid pace. Of course, the need for a cautious approach, as people’s health is at stake, no doubt plays a role, but there is more to it than that.
Challenges in using AI in Healthcare
Knowledge and expertise exchange is essential in enabling the application of AI techniques within the health domain. Without AI specialists and medical experts sharing knowledge, systems can be developed without practical use or AI techniques could be used ineffectively.
Hence, there is a need for more interdisciplinary teams.
Furthermore, while huge advances are seen in the development of AI techniques, the characteristics of the medical domain often require more specialized techniques. This means that solutions need to be more tailored, focusing on safety and explainability, as well as using the vast body of domain knowledge and the interaction between AI based systems and doctors/patients. Such techniques should be developed further.
The heavy data load
AI techniques require substantial amounts of data to train on. Looking at datasets of individual hospitals is often not enough, especially not when considering rare diseases. Additionally, hospital data is only one aspect of someone’s health. It must be combined with lots of other data sources to compile a more complete picture of the patient. There are of course many ongoing initiatives to combine health data in multiple ways taking the sensitive nature of the data into account. These initiatives need to be developed further and become better available for AI & Health projects.
Is AI the right choice?
Another cause for the limited application in real care settings includes the need for extensive evaluation of the contribution AI can make. How would the quality of care differ should a physician be advised by AI or not? Having more solid evidence on the usefulness of systems could certainly accelerate the developments of AI in the health domain, and also drive the willingness to conduct more of such extensive evaluations.
When AI techniques are integrated into care settings this changes the way people work, how patients are treated, and perhaps also the business models of the care organizations or companies. Understanding the impact and also ways to implement these is crucial for successful applications, and more research should be conducted to get a better understanding of that process.
Finally, once results show AI techniques can contribute in particular health settings, the approaches should also be brought to the market. For this, companies or startups with expertise in developing software for health settings come into play. We see increasing activities in that area.
What is Amsterdam doing?
Amsterdam has an excellent ecosystem in AI and Data Science, as well as world-leading health research institutes, medical centers, and companies. This places the city in a unique position to address the challenges listed above. Under the “AI Technology for People” initiative, knowledge institutes, including medical centres, in Amsterdam are collaborating on various themes including AI & Health, Fig. 1.
One Amsterdam wide initiative is Amsterdam Data Science (ADS) which brings together parties interested in Data Science and AI, independently of the application domain. As part of the ADS ecosystem, Amsterdam Medical Data Science (AMDS) has been established to bring together medical professionals and technical experts to further advance the application of AI in the field. This includes monthly meetups with presentations from medical professionals and technical experts, networking opportunities, as well as some seed funding. Exchanging knowledge, fueling interdisciplinary collaborations and learning from each other’s experience are AMDS’s main goals.
In collaboration with the Vrije Universiteit (VU) Amsterdam, Amsterdam UMC – location VUmc and ACTA, the VU Campus Center for AI & Health has been launched. This will act as a networking organization on the VU Campus, meant to fuel collaborations, make researchers aware of ongoing projects, but also showcase the developments of AI methods on the Campus. The ultimate goal is to improve healthcare by developing, implementing and evaluating AI technologies. Over 70 permanent staff members and dozens of postdocs and PhD students across nearly all VU faculties participate in the initiative.
If you’re interested in collaborating with AMDS or The VU Campus Centre for AI & Health please email email@example.com.
Managing/Being a Master’s Student during a Pandemic
For the past five years, Elsevier has been an enthusiastic participant in the UvA Master’s Student programme. In total, more than 45 students have been supervised by researchers across the company, which has led to 12 new recruits for our Data Science teams.
Data as a material for fashion: How treating data as a material enables a new future for design
Data Science is rapidly changing industries around the world, yet the digital transformation remains difficult for Fashion. Fashion (Design, Business, Branding, and Marketing) has never been known for maths geniuses. (There are a few, but they keep it a secret.) While maths and data may not be a given in the industry, people who work in fashion are material experts. So what would it mean if we treated data as if it were a material?
Programming Training for Refugees
In August 2020, VodafoneZiggo and Accenture wrapped up their three-month CodeMasters training programme for refugees. The training course was tailor-made to help refugees integrate in the Dutch labour market by teaching participants to write computer code.