A prospective study of 65 preoperative patients demonstrated that handheld echocardiography equipped with artificial intelligence software can comprehensively assess left ventricular diastolic dysfunction with accuracy comparable to standard cart-based systems, according to research presented at the American Society of Echocardiography (ASE) 2025 Scientific Sessions.

Researchers evaluated the Echonous handheld device (Kosmos tablet) with US2.AI software against traditional cart-based echocardiography, measuring all parameters required by ASE and British Society of Echocardiography (BSE) guidelines for left ventricular diastolic dysfunction (LVDD) classification. The randomized comparison included ejection fraction (EF), tissue Doppler velocities (septal and lateral e′), mitral inflow (E and A), E/e′ ratio, tricuspid regurgitation velocity, indexed left atrial volume (LAVi), and left atrial strain.

Correlation coefficients ranged from 0.72 to 0.98. Linear regression showed strong correlations across variables: average e′ r=0.97, E/e′ ratio r=0.95, TR velocity r=0.78, E/A ratio r=0.97, LA volume r=0.92, LA strain r=0.82, and LVEF r=0.91. Bland-Altman analyses demonstrated small mean differences with acceptable limits of agreement.

For diastolic dysfunction grading, there was statistically significant agreement between the AI-enabled handheld and cart-based measurements using both systems: Cohen’s κ=0.98 (ASE) and 0.93 (BSE), indicating near-perfect agreement.

Examinations were performed in randomized order. Handheld studies were uploaded to the US2.AI platform for automated analysis, while cart-based measurements were performed by clinicians. The study was a prospective observational investigation conducted with IRB approval and registered with the Clinical Trials Registry of India (CTRI/2024/10/075737).