Heart failure & cardiomyopathiesSecondary analysis
Comparison of Pulmonary Congestion Severity Using Artificial Intelligence-Assisted Scoring Versus Clinical Experts: A Secondary Analysis of BLUSHED-AHF
- Secondary analysis of the BLUSHED-AHF study
Authors' conclusion
Artificial intelligence/machine learning-based LCS correlated with expert-level B-line quantification.
Study context
- Population
- Patients with acute decompensated heart failure (ADHF) in the BLUSHED-AHF study of lung ultrasound (LUS)-guided therapy3,858 LUS clips
- Participants
- 130 patients
- Comparison
- AI/ML lung congestion score vs expert and ultrasound operator B-line quantification
- Primary endpoint
- Spearman correlation between AI/ML LCS and B-line counts from each of the three raters
Primary result
As reported in the abstract
Agreement of AI/ML LCS with expert B-line quantification (Spearman correlation)
Lung congestion score (LCS) vs expert A B-line score, Spearman r
0.894
LCS vs expert B B-line score, Spearman r
0.882
Both experts' scores agreed significantly better with LCS than with the ultrasound operator's score
p < 0.005, p < 0.001
Further findings
Experts' agreement with LCS vs with ultrasound operator's score
- Both experts agreed significantly better with LCS than with ultrasound operator's score
- P < 0.005, p < 0.001
Original publication
Comparison of Pulmonary Congestion Severity Using Artificial Intelligence-Assisted Scoring Versus Clinical Experts: A Secondary Analysis of BLUSHED-AHF. Eur J Heart Fail. Published 20 June 2023. doi:10.1002/ejhf.2881
Summarised from the published abstract; every number is checked against it