ICU sedation, delirium & recovery
Deep Learning–Based Recurrent Delirium Prediction in Critically Ill Patients
Authors' conclusion
Our delirium prediction model achieved strong performance by applying deep learning to a dataset that is at least one order of magnitude larger than those used in previous studies.
Study context
- Population
- Adult ICU admissions in 15 medical-surgical ICUs, Alberta, Canada, 2014–2020
- Participants
- 38,426 participants · 43,510 ICU admissions
- Outcome
- Prediction of delirium episodes in the next 12-hour periods (0-12 and 12-24 hr), from 24 hr after ICU admission
Methodological detailsprediction model development, deep learning, multicentre
- Prediction model development
- Deep learning
- Multicentre
Main finding
As reported in the abstract
Delirium prediction, 0-12-hour horizon (hold-out test set) · 0-12 hr
| Group | Sensitivity | Specificity | Precision (PPV) | AUROC |
|---|---|---|---|---|
| GRU model | 0.810 | 0.848 | 0.704 | 0.909 |
GRU model
Sensitivity0.810
Specificity0.848
Precision (PPV)0.704
AUROC0.909
Observational data: associations, not effects.
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Deep Learning–Based Recurrent Delirium Prediction in Critically Ill Patients. Crit Care Med. Published 15 February 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →