Us2.ai is proud to have been featured in seven scientific presentations at ESC Congress 2026 in Munich, Germany, including two late-breaking clinical science presentations. The research spanned some of the most clinically significant areas in cardiovascular imaging, from prognostic risk stratification and treatment monitoring in cardiac amyloidosis, to AI-assisted surveillance in cardio-oncology, to fully automated detection of pulmonary hypertension and hypertrophic cardiomyopathy from routine echocardiography.
The breadth of work presented at ESC 2026 reflects the growing role of Us2.ai across the full spectrum of cardiovascular disease, in specialist centres and community settings alike, and across populations from cancer patients to those with rare inherited cardiomyopathies.
Abstracts featuring Us2.ai:
Prognostic value of artificial intelligence-derived echocardiographic measurements in cardiac amyloidosis — demonstrating that Us2.ai-derived echocardiographic measurements provide robust prognostic stratification in cardiac amyloidosis, with performance comparable to manual measurements and incremental value over biomarker-based staging. Read more →
Changes over time in AI-quantified echocardiographic metrics before and after therapy in ATTR cardiomyopathy — showing that Us2.ai-powered echocardiography captures early attenuation of disease progression following disease-modifying therapy, supporting its use as a sensitive tool to monitor treatment response in ATTR-CM. Read more →
Validation of a deep learning workflow for echocardiographic interpretation in a cardio-oncology population — validating Us2.ai in 384 patients from a dedicated cardio-oncology centre at Royal Brompton Hospital, demonstrating strong LVEF agreement, improved interstudy repeatability, and a 43% reduction in study completion time. Read more →
Agreement between manual and AI-enhanced echocardiographic assessment of cancer therapy-related cardiac dysfunction in patients with early breast cancer: data from the CARE study — demonstrating that Us2.ai reliably reproduces longitudinal LVEF and GLS trajectories and rules out CTRCD in a large real-world breast cancer cohort, with a negative predictive value of 91%. Read more →
Screening performance of a deep learning model versus an automated wall thickness score in biopsy-proven cardiac amyloidosis — showing that the Us2.ai deep learning model achieved 98% sensitivity and a c-index of 0.82 in a biopsy-proven cohort, substantially outperforming an automated wall thickness score and supporting its use as a practical cardiac amyloidosis screening tool. Read more →
Late Breaking Presentations featuring Us2.ai:
Automated detection of pulmonary hypertension from a single apical four-chamber echocardiographic clip: development on the largest RHC-confirmed cohort to date and multinational external validation — a late-breaking clinical science presentation demonstrating that a spatiotemporal transformer trained on 17,969 RHC-confirmed patients can detect pulmonary hypertension from a single standard echo clip with no Doppler or multi-view input, achieving AUROC of up to 0.917 on external validation across three continents. Read more →
Deep learning model for detection, phenotyping, and left ventricular gradient estimation in hypertrophic cardiomyopathy: development and multinational external validation — a late-breaking clinical science presentation demonstrating that a fully automated end-to-end deep learning system detects HCM from routine echocardiography with 87.4% external sensitivity, exceeding the published expert benchmark of 50.7% by 37 percentage points, while also discriminating phenocopies and quantifying LVOT obstruction. Read more →