Heart failure & cardiomyopathiesDiagnostic or prognostic study
2-Dimensional Echocardiographic Global Longitudinal Strain With Artificial Intelligence Using Open Data From a UK-Wide Collaborative
- Diagnostic accuracy study
- external validation
- multiexpert consensus reference
Authors' conclusion
Our open-source approach to calculating GLS agrees with experts' consensus as strongly as the individual expert measurements and proprietary machine solutions.
Study context
- Population
- Apical 4-, 2- and 3-chamber echocardiographic images; external validation on 100 echocardiograms (Unity UK Echocardiography AI Collaborative)
- Participants
- 100 echocardiograms
- Comparison
- Open machine-learning global longitudinal strain (GLS) (Unity-GLS) vs individual experts and 2 proprietary packages
- Outcome
- Agreement of GLS with multiexpert consensus measurement
Performance
As reported in the abstract
Agreement with expert consensus GLS
| Test or model | Median absolute error | Correlation with consensus | Error |
|---|---|---|---|
| Individual experts | 2.00 GLS units | 0.85 | |
| Unity-GLS | 0.91 | 1.3 | |
| Proprietary A | 0.73 | 2.5 | |
| Proprietary B | 0.79 | 2.2 |
Further findings
| Outcome | Individual experts | Unity-GLS | Proprietary A | Proprietary B |
|---|---|---|---|---|
| Median GLS across 11 experts | median −16.1 (IQR −19.3–-12.5) | Not reported in the abstract | Not reported in the abstract | Not reported in the abstract |
| Rank correlation with multiexpert visual ranking | median rank correlation 0.72 | 0.77 | 0.70 | 0.74 |
Original publication
2-Dimensional Echocardiographic Global Longitudinal Strain With Artificial Intelligence Using Open Data From a UK-Wide Collaborative. JACC Cardiovasc Imaging. Published 10 July 2024. doi:10.1016/j.jcmg.2024.04.017
Summarised from the published abstract; every number is checked against it