Katarina at the Health Data Research UK Conference

Katarina presented her poster at the Health Data Research UK Conference in Glasgow, held 15–16 October. The conference brought together researchers, clinicians, policymakers and industry partners to discuss how health data and AI can be used to advance healthcare and improve patient outcomes.

Her poster, titled “GI-Clust: Deep Clustering for Early Gastrointestinal Cancer Detection”, introduces a deep clustering model designed to use UK primary-care Electronic Health Records (EHRs) to both identify patient phenotypes and classify gastrointestinal cancer subtypes. In this work, she uses longitudinal data from more than 210,000 patients within the QResearch database, one of the largest primary-care datasets in the UK.

A key feature of the model is an integrated interpretability framework, which provides clearer insight into the temporal clinical patterns underlying the discovered patient groups. This research is part of a wider collaboration between Oxford University, Queen Mary University of London, and Manchester University.