Insights from the first SEARCH industry-academia workshop
More than 20 experts from the SEARCH project and other initiatives participated in the first SEARCH industry–academia exploratory workshop, held online on 24 April. The session was organised within Task 5.1 to discuss recent developments in AI/ML for healthcare, identify possible areas of interest for future services, and gather initial input on how safe pre-competitive collaboration around the project could work in practice.
The workshop started with an introduction to the SEARCH project context, including synthetic data generation, synthetic data anonymity and credibility assessment, clinical decision support systems, and privacy preserving and federated collaboration approaches.
Synthetic data, trust, and validation
The discussion focused primarily on participant needs, relevance signals, and emerging developments. Participants agreed that synthetic data is seen as highly relevant for addressing key barriers in healthcare AI development, particularly limited access to real world data, scarcity of rare or specialised cases, long data access procedures, costly annotation, and challenges in multi site or multimodal data use. Participants also pointed to its potential to accelerate early research, enable proof of concept work, test algorithms under controlled or extreme conditions, support training and education, and contribute to product verification and validation.
A central finding was that the usefulness of synthetic data depends strongly on trust, validation, and context-specific credibility. Participants repeatedly stressed the need to assess whether synthetic data is realistic, privacy-preserving, fit for purpose, and suitable for downstream use in AI/ML development or clinical decision support. This suggests that validation and credibility support, documentation and provenance, benchmarking and evaluation frameworks, and federated collaboration support as promising candidate service interest areas for SEARCH.
The workshop also generated an initial watchlist of relevant developments, including multimodal foundation models, vision-language models, agentic AI, synthetic patient-history generation, green/efficient AI, and pan-European federated infrastructures.
Towards the SEARCH open innovation model
The workshop provided a valuable stakeholder informed basis for the further development of the WP5 open innovation model. Future activities will focus on refining potential service areas and advancing the open innovation approach through follow up actions, such as a second workshop, targeted surveys, and focused consultations with project partners and external stakeholders.
For more details, check out the the main conclusions of the session on this link.
New: SEARCH follow-up survey
To continue gathering stakeholder perspectives, the SEARCH project has also launched a survey focused on synthetic data, trust, validation, and future collaboration needs in healthcare AI. The survey explores topics discussed during the workshop, including the relevance of synthetic data for AI/ML development, validation and credibility requirements, benchmarking approaches, federated collaboration, and emerging developments such as multimodal foundation models, vision-language models, and agentic AI. Stakeholders from healthcare, research, industry, and innovation ecosystems are encouraged to participate and contribute to shaping the future SEARCH open innovation model.