Deep Dive: A Pilot Feasibility Study of AI-Generated Podcasts for Patient Education in Plastic Surgery

Background: Podcasts offer an accessible health communication medium. Generative AI allows rapid creation of high-quality, natural-sounding audio, presenting an opportunity for patient education. However, the feasibility and quality of AI-generated podcasts (AIGPs) for plastic surgery patient education are unexplored. Our objective was to assess the quality, accuracy, and effectiveness of AIGPs for patient education via plastic surgery provider evaluation. Methods: We developed ten educational plastic surgery podcasts using Google's NotebookLM. The model was grounded in institutional patient education materials, generated concise, formal, and informative episodes for patients. Twelve providers affiliated with a tertiary academic medical center's Division of Plastic Surgery evaluated the AIGP using the validated Questionnaire for Assessing Educational Podcasts (QAEP). Results: AIGP received high ratings across all evaluation domains (overall mean Likert = 4.69 ± 0.55). Technical audio quality achieved perfect scores (5.00 ± 0.00). Design and Structure (4.88 ± 0.34) and Content Adequacy (topic coverage, relevance, and appropriateness) (4.83 ± 0.38) were top-rated categories. Participants agreed podcasts were suitable for patient education and improved understanding (4.72 ± 0.52). Style and Syntax (4.44 ± 0.77) and Tone and Appropriateness (4.58 ± 0.67) showed highest variability. Conclusions: AIGP are a feasible and accurate patient education tool in plastic surgery, highly rated for technical quality, clarity, and educational value. Source-grounded generative AI can produce engaging, human-like audio with minimal human input. Further research should focus on patient comprehension and satisfaction to guide safe and scalable integration of AI-generated podcasts in surgical education.