AI-Based Quality Assessment of Ultrasound-Guided Upper Extremity Nerve Block Videos on YouTube: A Comparison of Speech-Text-Only and Image-Enhanced Approaches

Abstract

Objective: Ultrasound-guided upper extremity nerve block procedures require advanced visuomotor coordination. Although YouTube is widely used as an educational resource in this field, the educational quality of available content remains highly heterogeneous. This study aimed to compare the effectiveness of artificial intelligence (AI)–based quality assessment using a speech-text-only evaluation (Group 1) versus an image-enhanced evaluation (Group 2) for ultrasound-guided upper extremity nerve block educational videos. Material and Methods: In this descriptive cross-sectional study, a total of 80 ultrasound-guided upper extremity nerve block educational videos selected from YouTube on December 10, 2025 were analysed. Each video was evaluated under two conditions using ChatGPT-5.2 (OpenAI). Educational quality was assessed using a 10-item Educational Content Checklist, the DISCERN instrument, the Global Quality Scale (GQS), and the Journal of the American Medical Association (JAMA) benchmark criteria.

Results:  Group 2 demonstrated statistically significantly higher scores than Group 1 across all quality assessment tools: Educational Content Score (33.90 ± 6.25 vs. 29.11 ± 6.66; p < 0.001), DISCERN (45.01 ± 10.83 vs. 37.90 ± 8.62; p < 0.001), GQS [median 4 (4–5) vs. 3 (3–4); p < 0.001], and JAMA [median 2.5 (2–3) vs. 0 (0–0); p < 0.001].

Discussion and Conclusion: The image-enhanced AI approach yielded higher quality scores than the speech-text-only evaluation, particularly for procedurally dependent items reliant on visual cues. These findings suggest that speech-text-only evaluations may systematically fail to capture critical procedural elements in ultrasound-guided nerve block education.