Research exploring the use of artificial intelligence (AI) in automating echocardiographic measurements, the impact of baseline left ventricular (LV) health on the benefit of early TAVR, and aortic regurgitation (AR) following use of percutaneous transaortic LV assist devices (pLVAD) was presented at ASE 2025 from Sept. 5-7 and simultaneously published in JACC and JACC: Cardiovascular Interventions.

The first study found that an open-source deep learning model – EchoNet-Measurements – shows promise in aiding clinicians by automating echocardiographic quantification with a high degree of accuracy.

Yuki Sahashi, MD, MSc, PhD, et al., used 877,983 echocardiographic measurements from 155,215 studies from Cedars-Sinai Medical Center (CSMC) taken between 2011 and 2023 to train the deep-learning semantic segmentation models designed to automate the measurement of 18 echocardiographic parameters. Model outputs were validated by comparing sonographer measurements of temporal split data from CSMC as well as with an external data set from Stanford Healthcare.

The models showed high accuracy across all nine B-mode and nine Doppler measurements (mean coverage probability of 0.796 and 0.839 and mean relative difference of 0.120 and 0.096 on held-out data from CSMC and external data from Stanford Healthcare, respectively). Investigators evaluated 2,103 distinct studies from CSMC end-to-end, finding that the EchoNet-Measurements demonstrated similar reasonable performance (mean coverage probability 0.803 and mean relative difference 0.108).

Model performance was consistent regardless of patient age, sex, atrial fibrillation, obesity status and machine vendors.

"We have released the code and weights of these models publicly, along with demo user interface, to facilitate further research and clinical deployment in the field," write the authors.

In an accompanying editorial comment, Timothy J. Poterucha, MD, FACC, et al., state that "AI systems like the one described here may help address [automation bias] by providing verifiable measurements in contrast to less understandable outputs of other techniques, potentially enabling us to both see and believe."

Another study found that patients with asymptomatic severe aortic stenosis (AS) saw a consistent benefit with early TAVR treatment, regardless of baseline LV health.

This analysis of the EARLY TAVR trial included 901 patients randomized to either TAVR or clinical surveillance at 75 centers in the U.S. and Canada. Brian R. Lindman, MD, MSc, FACC, et al., defined integrated LV health as a composite of 1) absolute LV global longitudinal strain ≥15%; 2) LV mass index <115 g/m² (men) or <95 g/m² (women); 3) left atrial volume index ≤34 mL/m².

In the intent-to-treat cohort, 27% of participants had normal integrated LV health. Meanwhile, abnormal LV health was associated with higher event rates. When comparing clinical surveillance patients with those in the valve implant cohort, the benefit of early TAVR was seen regardless of LV health status.

Additionally, those in the clinical surveillance group were less likely to have normal LV health at two years vs. those treated early (36% vs. 48%, p<0.001). Although LV health tended to decrease in the clinical surveillance cohort from randomization to preprocedure, LV health at randomization was not predictive of the timing of aortic valve replacement or the severity of presentation.

According to the authors, these findings suggest that using abnormal measures of LV health as a rationale to consider early intervention in asymptomatic patients is "not useful for identifying patients who will derive greater benefit from an early intervention strategy. Instead, all patients – including those with abnormal LV health and normal LV health – appear to benefit from an early intervention strategy."

An additional study determined that AR after pLVAD use is uncommon and often mild. Jad Zeitoun, MD, et al., included 98 patients who received pLVAD support at a single site between 2014 and 2024. Patients underwent transthoracic echocardiography both before pLVAD and after removal.

Before pLVAD use, 69% of patients had no AR, 30% had mild AR and 1% had moderate AR. After removal, 13% had new or worsened AR – 10% of whom developed mild and 3% of whom developed moderate or moderate-severe AR.

Overall, larger devices (Impella 5.0 and Impella 5.5) were used in 33% of patients and median support duration was seven days. Results showed that patients with new or progressed AR had longer support duration (median 12 vs. 6 days; p=0.032), and for those with moderate or greater AR, support was even longer (median 51 days; p=0.046). All those with moderate or greater AR had large-bore devices vs. 31% of others (p=0.033). Based on these findings, the authors write that "risk appears related to device size and duration."