Toward generalizable and interpretable machine learning models in healthcare: Insights from ICU outcome predictions

Lasse Bohlen, Julian Rosenberger, Nico Hambauer, Daniel Zähringer, Volkmar Franz, Patrick Zschech, Mathias Kraus · 2026

Read Toward generalizable and interpretable machine learning models in healthcare: Insights from ICU outcome predictions by Lasse Bohlen, Julian Rosenberger, Nico Hambauer, Daniel Zähringer, Volkmar Franz, Patrick Zschech, Mathias Kraus (2026) 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/PMC13253713.txt.

Description

Lasse Bohlen and colleagues test machine learning for intensive care in a 2026 study across two health systems. Models were built with data from a German university hospital and the MIMIC-IV database from the United States. The tasks were to predict death, length of stay, and return to hospital for ICU patients.

Models trained on local data beat models brought in from outside when enough local data existed. Simpler and mid-level models, such as additive models, traveled better across hospitals than complex ones. They also needed far less local data, and tools that explain model choices helped show where data gaps hurt results.

Catalog record

CREATORLasse Bohlen, Julian Rosenberger, Nico Hambauer, Daniel Zähringer, Volkmar Franz, Patrick Zschech, Mathias Kraus
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE127296
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13253713

Browse the full library · About the Tower of Babel