Katarina Showcases AI for ICU Research at NeurIPS 2024 with Support from WiML

In December 2024, Katarina attended NeurIPS in Vancouver, supported by a travel stipend from Women in Machine Learning (WiML). As part of the WiML Workshop, she presented her research on CAMELOT++, a deep clustering framework for multi-task learning on temporal electronic health record (EHR) data in ICU settings. Her work extended an existing state-of-the-art model to handle mixed data types by designing an encoder that separately processes static and dynamic features, improving both clustering quality and predictive accuracy. Building on this, she introduced a multi-task extension that integrates survival analysis into the clustering process, enabling the discovery of patient subgroups with distinct survival patterns and clinical outcomes. 

The workshop provided a valuable opportunity for Katarina to share her findings and engage in deep, insightful discussions with fellow researchers. She also volunteered with WiML, which allowed her to get to know the amazing WiML community and connect directly with industry professionals through the career fair.

During the WiML Reception Dinner, Katarina had the chance to hear from two inspiring speakers: Tatiana Chavdarova, a visiting professor at the Department of Electronics, Information, and Bioengineering (DEIB) at Politecnico di Milano, and Akiko Eriguchi, a Senior Researcher at Microsoft. Both shared powerful reflections on their journeys in machine learning, offering valuable perspective and encouragement to attendees at various stages of their careers.