Developing an explainable machine learning model to predict the mechanical ventilation duration of patients with ARDS in intensive care units.
Zichen Wang, Luming Zhang, Tao Huang and 5 others
PMID 36423504WHAT IT FOUND
A machine model estimated how long ventilated ARDS patients might need ventilation.
Its external predictions were off by 5.57 and 5.46 days, so it is not ready to change ICU care without prospective testing.
Key findings
01The study analysed 1,148 ARDS patients from MIMIC-IV for training and 1,697 from eICU-CRD and 29 from AmsterdamUMCdb for external testing.
02The XGB model had the most balanced external prediction performance, with root-mean-square error 5.57 and 5.46 in eICU-CRD and AmsterdamUMCdb.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study used existing ICU database records rather than following new patients prospectively. The AmsterdamUMCdb external test included only 29 ARDS patients, which the authors said weakens the testing power. The three databases differed in patient characteristics and mechanical ventilation duration, so results may not transfer easily to every ICU. Comorbidities were not included because of database limitations. Features were only collected at the start of mechanical ventilation, not after ventilation began.
Declared interests
The study was supported by Guangdong Provincial Key Laboratory of Traditional Chinese Medicine Informatization (2021B1212040007), and the authors reported no conflicts of interest.
The easy way to misread this
Do not read the model as ready to guide ventilation decisions. It was built from retrospective ICU records, the external root-mean-square errors were 5.57 and 5.46, and one external dataset had only 29 patients, so it has not been shown to improve patient care.