Providence, Rhode Island, May 27, 2026 /PRNewswire/ — In a new feasibility study published in the Annals of Family Medicine, general practitioners (GPs) in China were able to accurately detect carotid artery plaque (an indicator of cardiovascular disease) in high-risk patients using AI-assisted portable ultrasound. However, the authors caution that more extensive testing is needed before this approach can be widely adopted, as diagnostic performance was inconsistent across participating general practitioners.
Carotid plaques are accumulations in the arterial walls of the neck that are associated with the development of atherosclerotic cardiovascular disease. Point-of-care ultrasound, also known as POCUS, is a new tool used in primary care. Artificial intelligence (AI) capabilities have recently been added to POCUS to improve image quality and diagnostic accuracy. This study investigated whether AI-assisted POCUS, combined with a training program, could help general practitioners in China use POCUS to screen for carotid plaque in their community clinics.
This study was conducted in Shanghai, China. Seven GPs completed a training program that combined classroom instruction with hands-on practice using AI-assisted POCUS. They recruited 169 patients at high risk for atherosclerotic cardiovascular disease from four community health centers during routine outpatient visits and offered free carotid plaque testing. Each patient was scanned by both a general practitioner and a senior ultrasound specialist. The device uses AI software to analyze ultrasound images in real time and alerts the operator with an on-screen marker when it detects potential plaque. Two senior ultrasound specialists independently reviewed all archived records and images to determine who had carotid plaque. This served as a benchmark for evaluating GP findings.
When examining the presence or absence of plaque across patients, general practitioners correctly identified about 87% of patients with plaque and correctly excluded about 91% of patients without plaque. Their findings showed a high level of agreement with senior ultrasound professional benchmarks. Among vessels in which plaque was confirmed, missed cases were concentrated at the bifurcation point where the carotid artery bifurcates (18.6%, 22/118). The authors suggest that this may be due to limitations of the AI systems themselves in the field, or gaps in how GPs incorporate AI feedback into their workflows. Misidentifying plaques in patients who actually did not have them was less common.
Each scan took about eight minutes to complete, but the researchers acknowledged that this may be difficult to extrapolate to typical GP appointments, where consultation times tend to be short. They suggest that this approach may be suitable, at least initially, for more focused settings such as stroke screening clinics, community health outreach campaigns, and home care programs. The authors describe this as the first study of its kind to test structured POCUS training for Chinese general practitioners in a real clinical setting, including AI assistance.
