Research Visit to the Cambridge Centre for AI in Medicine
Three months at CCAIM working on generative models for synthetic tabular data
From February to April 2023 I was a visiting researcher at the Cambridge Centre for AI in Medicine (CCAIM), University of Cambridge.
The visit focused on generative models for synthetic tabular data — variational autoencoders, generative adversarial networks, and diffusion models. Medicine is a domain where the data that would most improve a model is often the data you are least able to share, and synthetic data offers a way around that constraint, provided the synthetic samples are faithful to the real distribution, diverse enough to cover it, and not simply memorised copies of the training set.
That tension between fidelity, diversity, and generalisation became a line of work that continued after the visit, including our comparative evaluation of generative models for synthetic tabular data presented at ICDSST 2025.
I am grateful to the CCAIM team for hosting me and for a genuinely productive three months.