ICU-TSB: A Benchmark for Temporal Patient Representation Learning for Unsupervised Stratification into Patient Cohorts

Dimitrios Proios, Alban Bornet, Anthony Yazdani, Jose F Rodrigues Jr, Douglas Teodoro{dimitrios.proios, alban.bornet, anthony.yazdani, douglas.teodoro}@unige.ch, [email protected] · 2025

Read ICU-TSB: A Benchmark for Temporal Patient Representation Learning for Unsupervised Stratification into Patient Cohorts by Dimitrios Proios, Alban Bornet, Anthony Yazdani, Jose F Rodrigues Jr, Douglas Teodoro{dimitrios.proios, alban.bornet, anthony.yazdani, douglas.teodoro}@unige.ch, [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-06192.txt.

Description

Proios and co-authors present a 2025 test set for sorting hospital patients into groups by how their illness unfolds over time. Records from intensive care units hold rich day by day data. Finding real patient groups in that data could help doctors match care to each person's path.

Their test set, called ICU-TSB, draws on three public data sets from intensive care. It scores methods by how well the groups found match known disease classes. The team tried plain number methods and two kinds of neural networks. The learned groups often matched real disease groups, though clear gaps remain for later work.

Catalog record

CREATORDimitrios Proios, Alban Bornet, Anthony Yazdani, Jose F Rodrigues Jr, Douglas Teodoro{dimitrios.proios, alban.bornet, anthony.yazdani, douglas.teodoro}@unige.ch, [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE23 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-06192

Browse the full library · About the Tower of Babel