Human and Artificial Intelligence (AI) Analysis of Patient Experiences of Periodontal Graft Surgery

Background/Objectives: The prominent role the internet plays in being a source of dental information prompts qualitative evaluation of relevant online content. This study aimed to explore patients' experience regarding periodontal graft surgery communicated through the social media platform YouTube. Methods: An initial YouTube search using the term "gum surgery experience" retrieved 40 videos. Graft surgery was the most frequently discussed procedure, and 19 relevant videos were included in the qualitative analysis. Video content was analysed using a combined human-centered and artificial intelligence (AI)-assisted approach. AI-supported analysis of viewer comments was conducted using ChatGPT-4 and Gemini-1.5 Pro. Themes generated by human and AI analyses were compared. Results: Nine key themes were identified from the 19 videos that satisfied selection criteria. Most themes were similar between human and AI analyses, with six overlapping and three unique. The most frequently coded theme was post-operative recovery (n = 177), with pain, work absence, eating difficulties, and disrupted oral hygiene commonly reported. Patient-clinician relationships were frequently highlighted, with mixed experiences regarding communication and trust. Positive experiences were reported more frequently than negative. Comment analysis revealed varied audience engagement and sentiments, emphasizing concerns about pain, recovery, and procedural anxiety. Conclusions: Key themes related to patient experiences were identified, notably concerns regarding post-operative recovery and patient-clinician relationships. Challenges in finding information prior to having surgeries motivated patients to provide support and advice on YouTube, emphasizing the need for patient-centered resources and effective patient-clinician communication. Integrating human and AI methods in qualitative analysis was efficient and insightful, with AI supplementing but not substituting human research.