Respiratory failure & ventilation
Automatic acute respiratory distress syndrome (ARDS) surveillance with chest X-ray recognition using convolutional neural networks
Authors' conclusion
A CNN-based deep learning model showed clinically significant performance, providing potential for faster ARDS identification.
Study context
- Population
- Adults admitted to a medical ICU, January 2003–November 2014; chest X-rays categorised as ARDS, Pneumonia or Normal
- Participants
- 15,899 adults
- Outcome
- Accuracy (sensitivity and specificity) of the deep learning model in differentiating chest X-rays between pneumonia, ARDS and normal lungs
Methodological detailstwo-step convolutional neural network pipeline
- Two-step convolutional neural network pipeline
Performance
As reported in the abstract
Distinguishing ARDS, pneumonia and normal patterns
| Test or model | Sensitivity | Specificity |
|---|---|---|
| Convolutional neural network (CNN) | 91.8% to 97.8% | 96.6% to 98.8% |
Convolutional neural network (CNN)
Sensitivity91.8% to 97.8%
Specificity96.6% to 98.8%
Further findings
| Outcome | Values |
|---|---|
| Validation in a previous ALI/ARDS dataset | CNN sensitivity 96.3% · specificity 96.6% |
Original publication
Automatic ARDS surveillance with chest X-ray recognition using convolutional neural networks. J Crit Care. Published 28 March 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →