Perioperative medicine & outcomes
Deep-learning-based postoperative desaturation prediction using spirometry image data
Authors' conclusion
These findings support the efficacy of both the deep learning model and the predictive index in forecasting postoperative desaturation
Study context
- Population
- Patients undergoing laparoscopic or robot-assisted abdominal surgery207 training, 45 external validation
- Participants
- 252 patients
- Comparison
- Deep learning model vs Desaturation predictive index
- Outcome
- Model's ability to predict postoperative desaturation (AUROC)
Methodological detailsmulticentre, convolutional neural network, Two tertiary hospitals for training; independent hospital for external validation
- Multicentre
- Convolutional neural network
- Two tertiary hospitals for training; independent hospital for external validation
Performance
As reported in the abstract
Prediction of postoperative desaturation
| Test or model | AUROC, internal validation (95% CI) | AUROC, external validation (95% CI) | AUROC, internal | AUROC, external |
|---|---|---|---|---|
| Deep learning model | 0.825 [0.744-0.905] | 0.688 [0.656-0.720] | — not stated in the abstract | — not stated in the abstract |
| Desaturation predictive index | — not stated in the abstract | — not stated in the abstract | 0.774 | 0.722 |
AUROC, internal validation (95% CI)
Deep learning model0.825 [0.744-0.905]
Desaturation predictive index— not stated in the abstract
AUROC, external validation (95% CI)
Deep learning model0.688 [0.656-0.720]
Desaturation predictive index— not stated in the abstract
AUROC, internal
Deep learning model— not stated in the abstract
Desaturation predictive index0.774
AUROC, external
Deep learning model— not stated in the abstract
Desaturation predictive index0.722
— not stated in the abstract
Original publication
Deep-learning-based postoperative desaturation prediction using spirometry image data. J Clin Monit Comput. Published 27 September 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →