AI-Generated text-to-video avatars: Enhancing education and engagement for postoperative instructions (Preprint)

Abstract

<sec> <title>BACKGROUND</title> Effective postoperative communication is vital for patient recovery, yet traditional text-based discharge instructions often lead to poor comprehension and adherence, particularly for patients with limited health literacy. While educational videos improve understanding and retention, their widespread use has been hampered by high production costs and logistical challenges. Generative AI offers a scalable solution to create personalized, engaging video content. </sec> <sec> <title>OBJECTIVE</title> This pilot study aimed to evaluate the effectiveness and user experience of AI-generated, avatar-led videos for delivering postoperative instructions compared to traditional text-based handouts. </sec> <sec> <title>METHODS</title> In this randomized pilot study, 38 healthcare worker volunteers were assigned to either a text-based instruction group (n=19) or an AI-generated video instruction group (n=19). Both groups received information on 10 common postoperative topics. Primary outcome was objective knowledge assessed via a 10-item quiz. Secondary outcomes, measured through surveys with 5-point Likert scales, included engagement time, subjective engagement, perceived clarity, usefulness, confidence in understanding, and information retention. Qualitative feedback was also collected. </sec> <sec> <title>RESULTS</title> Participants in the AI video group demonstrated significantly higher engagement times (mean: 15.11 ± 7.78 minutes vs. 8.84 ± 4.03 minutes for text; P = .0036, Cohen's d = 1.04). They also rated instructions as significantly clearer (4.63 ± 0.50 vs. 4.00 ± 0.82; P = .0066, Cohen's d = .93), more engaging (4.05 ± 0.78 vs. 3.32 ± 1.00; P = .0160, Cohen's d = .81), and more effective for information retention (4.42 ± 0.84 vs. 3.37 ± 0.68; P = .00015, Cohen's d = 1.38). However, objective knowledge quiz scores did not significantly differ between groups. Qualitative feedback highlighted the engaging and memorable nature of the AI videos but noted areas for avatar refinement (e.g., naturalness of speech and gestures). </sec> <sec> <title>CONCLUSIONS</title> AI-generated avatar videos show promise for significantly enhancing patient engagement and subjective understanding of postoperative instructions, despite no significant difference in objective knowledge scores in this pilot. While current AI avatar technology requires refinement, this scalable approach holds potential for creating more accessible and memorable patient education materials, potentially improving adherence and outcomes. Future research should explore larger patient populations, long-term outcomes, and hybrid text-video approaches. </sec>

Ähnliche Arbeiten