Working Group initiates Workshop on Small Data & AI in Medicine

Artificial intelligence (AI) has become an integral part of many aspects of everyday life. In medicine, AI can assist by identifying connections between symptoms and diseases from medical data and providing treatment recommendations. But what happens when only small data sets are available? Can AI still function effectively? These questions were discussed by thirty international researchers at the three-day scoping workshop "Data Augmentation and Imputation Methods for Health Data" in Hanover. This workshop, funded by the Volkswagen Foundation, was initiated by the working group “Small Data“ which is part of the UBRA Peer-to-Peer-network Artificial Intelligence.

By Nadine Metzler

28 Feb 2025

© Klaus Eickel

Scoping workshops funded by the Volkswagen Foundation provide researchers with the opportunity to explore new perspectives outside their ongoing research projects. These workshops bring together international experts to develop ideas for addressing identified research gaps and to engage in intensive discussions - an important opportunity for the field of medical informatics.

"When it comes to healthcare data, the potential of artificial intelligence for supporting diagnoses, predicting outcomes, or making treatment recommendations is very promising. However, a major challenge is that available data sets are often small or incomplete," explains Prof. Eickel. "One possible solution is data augmentation, where AI automatically fills in the gaps in the data."

But it’s not that simple. It must be ensured that AI-generated data does not distort the results. "We need to establish quality criteria and evaluate the impact of additional data on outcomes," says Prof. Eickel. Achieving this requires expertise from multiple disciplines.

During the scoping workshop, participants discussed what expectations should be placed on AI applications and which existing methods could be utilized. As a follow-up, they are now working on a position paper that is set to be published later this year.

Prof. Dr. Klaus Eickel conducts research at Fraunhofer MEVIS, focusing on the development of innovative, non-invasive biomarkers for diagnosing Alzheimer’s and multiple sclerosis, and is member of the UBRA working group “Small Data“. At Bremerhaven University of Applied Sciences, he teaches in the Medical Engineering program as a professor of medical informatics.