AI-Assisted Handheld Echocardiography by Nonexpert Operators: A Narrative Review of Prospective Studies

Handheld point-of-care ultrasound (POCUS) increasingly incorporates AI to assist nonexpert operators in echocardiography, guiding image acquisition and automating core measurements. This narrative review synthesizes recent (2018-2025) prospective studies evaluating AI-assisted handheld echocardiography performed by nonexpert users in adult, pediatric, ambulatory, and critical care settings. A structured PubMed/MEDLINE search was used to identify eligible studies that included prospective clinical trials and randomized controlled trials that assessed real-time AI guidance, image quality feedback, automated view classification, and/or automated measurements (e.g., ejection fraction (EF) analysis). Across nine studies, AI guidance consistently improved image acquisition, enabling nonexperts to obtain diagnostically adequate transthoracic views in most patients after limited training. In hospital and outpatient settings, diagnostic adequacy for key triage questions such as left ventricular function and pericardial effusion frequently exceeded 90%. Automated EF analysis achieved close agreement with reference echocardiography, while end-to-end nurse-led pathways demonstrated feasible integration into routine clinic workflows with short examination times. Educational trials further showed higher view-acquisition success, faster learning, and improved recognition of systolic dysfunction with minimal time penalties. Overall, AI-assisted handheld echocardiography supports reliable, triage-oriented cardiac imaging by nonexperts across diverse environments. Strengths include enhanced feasibility, consistency, and efficiency, while Doppler-dependent tasks show lower adequacy and should not be frontline targets currently. Broader validation, outcome-based evaluation, and structured governance are needed to ensure safe, equitable, and scalable implementation.