Respiratory failure & ventilation
Multicentre validation of a machine learning model for predicting respiratory failure after noncardiac surgery
Authors' conclusion
Utilising just eight easily extractable variables, this machine learning model demonstrated excellent discrimination in both internal and external validation for predicting postoperative respiratory failure.
Study context
- Population
- Noncardiac surgical cases (electronic health record (EHR) data)208,308 in external validation
- Participants
- 307,333 cases
- Comparison
- Derivation vs Validation cohort A vs Validation cohort B vs Validation cohort C
- Derivationn = 99,025
- Outcome
- Postoperative respiratory failure (prolonged (>48 h) mechanical ventilation or reintubation after surgery); model discrimination by AUROC and AUPRC
Methodological detailsmulticentre, external validation, gradient boosting machine learning
- Multicentre
- External validation
- Gradient boosting machine learning
Performance
As reported in the abstract
Discrimination for postoperative respiratory failure (AUROC, AUPRC)
- Internal validation area under the receiver operating characteristic (AUROC)
- 0.912 (95% CI 0.908-0.915)
- Internal validation area under the precision-recall curve (AUPRC)
- 0.113
- External validation cohort A AUROC
- 0.879 (95% CI 0.876-0.882)
- External validation cohort B AUROC
- 0.872 (95% CI 0.870-0.874)
- External validation cohort C AUROC
- 0.931 (95% CI 0.925-0.936)
- External validation cohort A AUPRC
- 0.029
- External validation cohort B AUPRC
- 0.083
- External validation cohort C AUPRC
- 0.124
Original publication
Multicentre validation of a machine learning model for predicting respiratory failure after noncardiac surgery. Br J Anaesth. Published 26 February 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →