Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes

The paper proposes a new method for predicting extubation failure (EF) by extracting features from free-text respiratory therapy notes using a large language model and combining them with logistic regression. Applied to a cohort from University of Washington Medicine, the approach identifies clinically relevant EF-related features that enhance prediction performance when added to structured patient data. The study also discusses how varying target populations and EF definitions in prior research can cause systematic performance differences and limit generalizability.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

• llms