Respiratory failure & ventilation
Prospective Real-Time Validation of a Lung Ultrasound Deep Learning Model in the ICU
Authors' conclusion
A previously validated DL classification model performs equally well in real-time at the bedside when platformed on a portable device.
Study context
- Population
- Critically ill ICU patients receiving oxygen therapy and eligible for respiratory imaging; unstable patients or those unable to undergo lung ultrasound (LUS) excluded400 clips
- Participants
- 100 patients
- Comparison
- Real-time deep learning (DL) model vs blinded expert annotation (gold standard) for A vs B line pattern on LUS
- Outcome
- Identification of B line pattern
Methodological detailsprospective, observational, real-time bedside deployment, blinded expert review, two tertiary-care sites, 56 patients mechanically ventilated
- Prospective
- Observational
- Real-time bedside deployment
- Blinded expert review
- Two tertiary-care sites
- 56 patients mechanically ventilated
Performance
As reported in the abstract
| Test or model | Accuracy | Sensitivity | Specificity |
|---|---|---|---|
| Real-time inference | 95% | 93% | 96% |
Real-time inference
Accuracy95%
Sensitivity93%
Specificity96%
Further findings
Varying prediction thresholds (4 approaches)
- Real-time modification of sensitivity and specificity according to clinical priorities is possible
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Prospective Real-Time Validation of a Lung Ultrasound Deep Learning Model in the ICU. Crit Care Med. Published 20 January 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →