Accurate assessment of left ventricular ejection fraction (LVEF) is essential for diagnosing and managing heart failure. Yet access to echocardiography in geriatric units remains limited, often requiring specialist cardiologists that are not readily available in these settings. This prospective study, conducted at Hospital Broca of Paris, evaluated the relevance and feasibility of AI-assisted automated LVEF measurement (AutoEF-AI), combining a handheld imaging device with the Us2.ai analysis software, in 136 patients hospitalised for acute heart failure in geriatric units. Each patient underwent both standard echocardiography by an expert cardiologist and AutoEF-AI echocardiography performed by a geriatrician who had completed just a single day of echo training.
Key findings
- Excellent agreement between AutoEF-AI and the expert reference method (ICC = 0.96, r = 0.97)
- 99% sensitivity and 89% specificity for detecting abnormal LVEF (<50%), with an overall diagnostic accuracy of 93%
- Minimal mean bias of 1.31%, with narrow limits of agreement
- The presence of a pacemaker was the only variable significantly associated with measurement discrepancy
Implications
These findings suggest that AI-assisted echocardiography with Us2.ai has strong potential to extend high-quality cardiac assessment beyond specialist settings. By enabling geriatricians with limited echo training to perform accurate LVEF measurements, the technology could help close the diagnostic gap in geriatric care, supporting earlier heart failure detection and more timely treatment decisions for a vulnerable patient population.