At the heart of the project is synthetic health data – artificially generated data designed to preserve important statistical properties of real clinical data while protecting patient privacy. The researchers will investigate a critical question: can synthetic data reproduce not only the characteristics of real patient populations, but also the causal relationships and treatment effects found in the original data.
Using health data from Sweden and South Korean, the collaboration will develop and evaluate new generative AI methods for epidemiological research, including survival and treatment-effect analyses. The researchers will also investigate whether models developed using Swedish data can support analyses of Korean populations without transferring raw patient information between countries.
The project has strong relevance for CIRCE, where researchers integrate large-scale clinical, genomic, registry, and population data to understand cancer risk, treatment outcomes, survivorship, and health inequalities. Privacy-preserving approaches could make it possible to validate findings across populations while maintaining robust protection of sensitive health information.
More broadly, the project brings together expertise in epidemiology, causal inference, and artificial intelligence, to develop new ways of conducting secure data-driven health research across national borders.
The collaboration between Lund University and Seoul National University is funded by the Swedish Research Council through its 2026 call for research collaboration between South Korea and Sweden.