Last week, a networking meeting between The Wallenberg AI Autonomous Systems and Software Program (WASP-HS), and SciLifeLab and Wallenberg program for Data-driven Life Science (DDLS) took place. The purpose of the meeting was to explore opportunities for joint and multidisciplinary research ventures between researchers from both programs, and a call for seed-money, meant to facilitate these collaborations, was announced.
The network sessions were divided into two parallel sessions with two different themes: Session 1 with Data challenges and Implementation and Use, and session 2 with Prediction and Modeling, and Governance and Economics. Each session had two moderators, one from each program.
Data Challenges
Moderator WASP-HS: Francis Lee, Associate Professor in Science and Technology
Moderator DDLS: Helga Westerlind, Assistant professor, Clinical Epidemiology Division, Karolinska Institute
The Data Challenge theme focuses on the ethical, organizational, and social challenges around data work, and addresses questions like; what are the challenges in creating representative datasets for diverse populations?
A few of the topics that were discussed during the session was; Agent Based Simulation, Explainability and how it ties into ethical use of AI, Single cell sequencing data, The Ideal data reusability system, Transparency and stakeholder views, and using Social media as proof. Is it ok to use public data? Practical, legal and ethical challenges surrounding data was also discussed.
”It was a very engaging and inspiring meeting, and a lot of fun to meet and network with other researchers that use similar methods, but in different fields. During the discussion on data challenges I could clearly see how collaborating with a researcher from WASP-HS would improve my own research, especially from the so very important patient perspective”, says Helga Westerlind, Assistant professor, Clinical Epidemiology Division. “It was also very obvious during the session that a lot of us struggle with similar, very practical, problems, including both infrastructure and legal issues.”
Implementation and Use
Moderator WASP-HS: Helena Lindgren. Professor in Computer Science, Umeå University
Moderator DDLS: Erik Kristiansson, Professor, Chalmers University of Technology
This theme addresses questions regarding the ethical, organizational, and social challenges around implementation and use. Questions could include for instance: How can the generated medical knowledge be implemented in medical education and clinical practice? And, what are the legal and ethical implications thereof?
During the session, the participants discussed the implementation and use of an AI-based decision support system in clinical routine. The possible advantage and disadvantage of such a system was also discussed, as well as how to use/re-use already existing data and knowledge. The participants also highlighted the importance of trust in the support system and that the system might be used for both education and training of professionals, but also to introduce the tools use when entering clinical practice.
”Interdisciplinary collaborations is fundamental for the success of the Data-Driven Life Science (DDLS) program. We are excited to see that there is a large and growing interest in many scientific areas connecting both DDLS and WASP-HS. Further promotion of research related to the many societal challenges and opportunities of artificial intelligence will be necessary to fully leverage of these new promising technologies”, says Erik Kristiansson.
Prediction and Modeling
Moderator WASP-HS: Francis Lee, Associate Professor in Science and Technology Studies, Chalmers University of Technology
Moderator DDLS: Erik Kristiansson, Professor, Chalmers University of Technology
This theme focuses on the ethical, organizational, and social challenges around prediction and modeling, and addresses questions such as; what are the ethical and social issues in modelling and forecasting?
During the session, the participants discussed the reuse of AI models and transfer learning, and how models trained in one place and used in another, can help spread knowledge, as well as provide opportunities and new perspectives. Various challenges were also discussed, including the need for new and improved methods that can identify which part of an AI algorithm fails, which in turn can lead to the identification of many new opportunities that are currently overlooked. Another problem highlighted, was the possible differences in ethical restrictions between various clinical settings or populations.
Another interesting topic, that could potentially open up for new discoveries, is the use of natural language AI models on DNA sequences. In these cases, the sequence is viewed as a language with its own grammar.
”The meeting was a great success in gathering researchers from different disciplines around a common set of problems. To me, the meeting brought up fundamentally important questions about our data driven future. I think working together in collaborations between computer scientists, life scientists, and social and humanistic researchers is central to understand and handle the challenges—in science, technology, society, and organizations—that a more data driven life science brings. This meeting was a really good starting place for these kinds of collaborations”, says Francis Lee.
Governance and Economics
Moderator WASP-HS: Liane Colonna, Assistant professor in law and information technology, Stockholm University
Moderator DDLS: Heidi Howard, Researcher, Medical Ethics, Lund University
This theme, centers on the legal, organizational, and governance challenges around data-driven life science. Questions could include for instance: Legal aspects of data (use, reuse, transfer, sharing etc.) What is/could be/should be the big picture (non-technical) model of data governance at DDLS?
In this session, the participants discussed the need for more events and activities where the humanities and social science researchers can meet and discuss with data scientists, but also that each researcher explains their work in a very basic way, to avoid losing important information “in translation”.
Another important question that came up was the need for innovation offices at universities and hospitals to participate in different pilot project introducing automated decision making in hospital care, together with health technology assessment groups. These kind of stakeholders should also be considered when trying to bring people together to discuss ethical, legal and social aspects of AI/data science use in health care.
“This DDLS and WASP-HS collaboration shows a true will, a concrete process, and funds for encouraging meaningful multidisciplinary research in Sweden. While “multi- and interdisciplinary research” are identified as goals of many research organizations and funders, there are also challenges to properly support such research. This includes aspects like “domain protectionism”, where funds keep going to more traditional research fields and questions, as well as the difficulties that come with trying to evaluate truly innovative and multidisciplinary projects. This collaboration is “walking the talk” so to speak, and is making meaningful multidisciplinary projects happen”, says Heidi Howard.
