Ketogenic diet: Assessing YouTube video information using quality, reliability, and text analytics methods

Abstract

Objective Patients and the general audience refer social media platforms, such as YouTube, to learn and apply contemporary dietary methods. It is difficult for users to analyze the correctness and quality of information available on open platforms. Using scientific evaluation, this study assessed the quality, reliability, and content of YouTube videos on ketogenic diet (KD). Methods Three experienced medical practitioners reviewed and evaluated 95 videos. The quality and reliability of the videos were assessed using the quality criteria for consumer health information and the global quality scale (GQS). Topic modeling and sentiment analysis were employed to determine the dominant themes and polarity of the information. Results Three types of publishers (doctors, educational institutions, and influencers) were identified for the study. The mean length of videos posted by doctors was high at 42.24 min. The reliability and quality scores ranging from 0 (low) to 5 (high) had an average of 3.08 ± 1.14 and 3.18 ± 1.18, respectively, for sampled videos. One-way analysis of variance reveals significant differences in DISCERN and GQS scores among doctors, educational institutions, and influencers. Topic discovery identified four themes: keto versus glucose, diabetes, KD food, and major chronic diseases. Sentiment analysis reveals positive content polarity, some content shared by doctors had a neutral sentiment. Conclusion Content creators should augment the content by citing medical information and terminology. Viewers relied more on doctors for information related to KD. The aesthetic quality is high for all types of publishers. Publishers could focus on the discovered themes to create more content. Publishers should produce high-quality videos by improving esthetics (to increase engagement), and reliable medical information (to increase impact).