Erschienen in:
05.11.2022 | Correspondence
Building a better machine learning model of extubation for neurocritical care patients
verfasst von:
Shohei Ono
Erschienen in:
Intensive Care Medicine
|
Ausgabe 1/2023
Einloggen, um Zugang zu erhalten
Excerpt
I read the article by Cinotti et al. [
1] with great interest and appreciate the value of the authors’ efforts to create a model for predicting extubation failure in neurocritical care patients with impaired consciousness. If this study allows us to predict extubation failure, we should be able to reduce the number of patients receiving unnecessary ventilator management, thereby decreasing the number of complications related to extubation failure. However, the area under the curve in this study was 0.71 in the validation cohort, which we would expect to be even lower if the external data had been validated. With its current predictive accuracy, this model may not be useful in clinical practice. We would like to point out four areas of concern regarding the machine learning method in this study. …