Respiratory failure & ventilation
Subphenotyping prone position responders with machine learning
Authors' conclusion
Distinct ARDS subphenotypes with varying mortality were identified in patients undergoing prone positioning; however, predicting which patients benefited from this intervention based on available data was not possible.
Study context
- Population
- Mechanically ventilated acute respiratory distress syndrome (ARDS) patients who underwent at least one prone positioning cycle
- Participants
- 353 participants
- Exposure
- ARDS subphenotypes identified by machine learning
- Outcome
- 28-day mortality
- Follow-up
- 28 days
Methodological detailsunsupervised machine learning
- Unsupervised machine learning
Main finding
As reported in the abstract
Cluster 3
56%
- Mortality differed significantly across clusters; Cluster 3 highest
- P = 0.03
Observational data: associations, not effects.
Secondary outcomes
Response to prone positioning (compliance, driving pressure, PaO2/FiO2, ventilatory ratio, mechanical power)
- No significant differences in proportions of responders for any parameter
Transpulmonary pressure (subcohort)
- Did not improve subphenotype characterization
Original publication
Subphenotyping prone position responders with machine learning. Crit Care. Published 14 March 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →