Respiratory failure & ventilation
Machine learning predicts lung recruitment in acute respiratory distress syndrome using single lung CT scan
Authors' conclusion
ML based on a single CT data at 5 cmH2O represented an easy-to-apply tool to classify ARDS patients in recruiters and non-recruiters according to both radiologically defined and gas exchange-defined lung recruitment within the first 48 h from the start of mechanical ventilation.
Study context
- Population
- Mechanically ventilated, sedated and paralysed patients with acute respiratory distress syndrome (ARDS)retrospectively enrolled
- Participants
- 221 participants
- Comparison
- CT-based machine learning (ML) vs Combined ML
- Outcome
- Classification of lung recruiters: radiologically defined (Δ45-5 non-aerated tissue > 15%) and gas exchange-defined (Δ15-5 PaO2 > 24 mmHg)
Methodological detailsretrospective, machine learning classifier development and validation, Four ML algorithms evaluated, PEEP trial at 5 and 15 cmH2O and two lung CT scans at 5 and 45 cmH2O airway pressure
- Retrospective
- Machine learning classifier development and validation
- Four ML algorithms evaluated
- PEEP trial at 5 and 15 cmH2O and two lung CT scans at 5 and 45 cmH2O airway pressure
Performance
As reported in the abstract
Classification of radiologically defined lung recruiters
- ML based on CT data at 5 cmH2O classified radiologically defined recruiters with similar AUC as ML based on combined lung mechanics, gas exchange and CT data
Further findings
Classification of gas exchange-defined lung recruiters
- ML based on CT data classified gas exchange-defined recruiters with the highest AUC
Original publication
Machine learning predicts lung recruitment in acute respiratory distress syndrome using single lung CT scan. Ann Intensive Care. Published 5 July 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →