AI-generated videos in medical science popularization and education: a scoping review

This scoping review aimed to systematically review research on the application of artificial intelligence-generated (AI-generated) videos for the general public education, patient education, medical student education, and physician education, and propose a framework covering the entire video production process.

Arksey and O'Malley's scoping review guidelines were followed, using the PRISMA-ScR checklist. Studies were identified from PubMed, Web of Science, China National Knowledge Infrastructure and Google Scholar using keywords and medical subject headings terms related to "AI-generated", "evaluation" and "video", and screened for relevance, with any duplicates removed.

The initial 1,206 articles were reduced to 10 for review. Of these, half failed to clearly document the prompts used. Those that did provided only brief descriptions based on the application context or keywords, and the AI-generated videos consistently received low scores on validated instruments such as the Global Quality Scale (GQS), with most scoring 2 out of 5, indicating limited suitability for patient education. Comprehensive quality assessments were rare: only one study used three distinct evaluation tools, and the rest used one or two or none at all, revealing significant shortcomings in the standardization and comprehensiveness of the overall evaluation framework.

To the best of our knowledge, this is among the first scoping reviews to synthesize the emerging evidence on AI-generated videos for the general public education, patient education, medical student education, and physician education. Few studies are reported and they demonstrate considerable heterogeneity. We identified insufficient standardization in prompts, generation pathways, tools, and evaluation instruments. We propose a procedural framework that may guide and inform future research in this field.