JISARA

Abstract: This study assesses the quality of nutrition-related shorts on YouTube, focusing on the prevalence of credible content. Using Google Trends, we analyzed YouTube search behaviors for key nutrition terms over five years. We found peak years before and after Covid. Based upon Google Trends peak years, we developed python code using an API key to fetch YouTube shorts. A total of 2,391 YouTube shorts were fetched and further assessed for quality inclusion. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework was utilized to illustrate the video selection process/parameters. Shorts were excluded based on duplicates, content, author channel credentials, duration, not in English, no words spoken or an ad. A remaining total of 30 videos were selected to be analyzed via Google AI Studio’s Gemini 2.5 Pro Preview 05-20 regarding their reliability, quality of information, and overall quality using the DISCERN instrument. Results revealed only 6 of the selected shorts had an overall high-quality score. While this study is limited to YouTube and nutrition-related content, its findings offer valuable insights for health professionals, policymakers, and educators aiming to improve digital health literacy. It also adds practical value to technology educators teaching students to fetch data via an API and assessing the quality of data via LLM.

Download this article: JISARA - V19 N4 Page 53.pdf

Recommended Citation: Powell, L.M., Powell, G., Rebman Jr., C.M., Wimmer, H., (2026). Using a Large Language Model to Evaluate Content Quality in Nutritional YouTube Shorts. Journal of Information Systems Applied Research and Analytics 19(4) pp 53-69. https://doi.org/10.62273/HFNS3900