This document discusses analyzing YouTube videos about diabetes as reliable health sources. It finds that while YouTube contains useful health information, some videos may contain misleading information. The document then explores using social features like likes, dislikes, comments and author reputation to identify reliable health videos. It searches for diabetes videos from organizations on a healthcare social media list and finds few in top results. A new approach is proposed that ranks videos from hospitals, organizations and active users higher to improve reliability of search results. Future work would analyze interactions and metadata of these videos and users.