User validation of an AI-assisted ultrasound guidance system for rapid vascular access

Vascular cannulation enables life-saving resuscitation in traumatic hemorrhage. The AI-GUIDE-LL is a handheld ultrasound-guided robotic device enabling inexperienced medical providers to obtain femoral vascular access. Device efficacy has been demonstrated in expert-operated porcine models, though performance among inexperienced operators and in humans is unknown. We evaluated the AI-GUIDE-LL's performance among novice clinicians to target human femoral vessels and accurately deploy needles in phantoms.

This single-site prospective study recruited medical professionals without ultrasound-guided procedural experience. Users attempted femoral artery needle insertion in five human CT-derived inguinal phantoms using fluid aspiration as the reference standard, with three needle insertions permitted per phantom. Subsequently, they used the device to target healthy human volunteer femoral vessels, assessing target vessel localization and common femoral segment identification. Success was defined by expert review of ultrasound images and video recordings.

Thirty clinicians (seven advanced practice providers, five EMTs, nine physicians, and nine registered nurses) achieved correct needle insertion in phantoms in 93% (140/150) of trials, with a median completion time of 44 s and a mean of 1.1 attempts. 28 volunteers were scanned by 28 clinicians. Clinicians successfully targeted 97% (93/96) of vessels with a median time of 25 s. Common femoral segment confirmation was achieved in only 20% of trials.

The AI-GUIDE-LL enables rapid, accurate femoral vessel localization and needle insertion among novice clinicians, supporting feasibility for field-forward military and point-of-care civilian settings. Reliable segment-specific localization above the femoral bifurcation, necessary for large-bore arterial procedures, remains an important limitation requiring further work.

Question AI-GUIDE-LL is a handheld robotic system that integrates artificial intelligence and ultrasound guidance to assist non-expert users in performing femoral vascular access for rapid point-of-care interventions. Findings Clinicians with limited ultrasound experience successfully localized femoral vessels, achieving 93% phantom needle insertion success and 97% target vessel localization in human trials. Clinical relevance AI-GUIDE-LL may enable clinicians with limited ultrasound training to perform timely femoral cannulation for life-saving interventions, including fluid resuscitation in traumatic hemorrhage, particularly in military or resource-limited environments.