%0 Journal Article
%T Investigation of Large Language Models’ Capabilities in Answering Clinical Questions Related to Periodontal Furcation Therapy
%A Dimitris Kounatidis
%A Bhanu Raghunathan
%J International Journal of Dental Research and Allied Sciences
%@ 3062-3502
%D 2025
%V 5
%N 2
%R 10.51847/pPUa3JWPaM
%P 1-9
%X Large Language Models (LLMs) represent advanced artificial-intelligence (AI) tools capable of processing massive textual datasets and producing language that resembles human expression. Their emergence suggests new possibilities for retrieving clinically relevant knowledge within healthcare. This investigation sought to measure and compare how four separate LLMs responded to practical questions on the management and therapy of periodontal furcation defects, assessing the degree to which their answers aligned with scientific evidence. Four models—ChatGPT 4.0, Google Gemini, Google Gemini Advanced, and Microsoft Copilot—were each prompted with ten clinical questions on furcation-defect management. Their replies were benchmarked against a reference source drawn from the European Federation of Periodontology (EFP) S3 guidelines and recent systematic reviews. Two certified periodontists independently graded every response for depth, factual correctness, clarity, and clinical appropriateness, following a fixed rubric that assigned 0–10 points per criterion. Performance varied among systems. Google Gemini Advanced consistently achieved the highest averages, especially for breadth and readability, whereas Google Gemini and Microsoft Copilot tended to receive lower marks, most notably in relevance. A Kruskal–Wallis comparison, however, detected no statistically significant overall difference in mean scores. Inter-rater and intra-rater reliability were both strong. Although all LLMs demonstrated an ability to generate answers regarding furcation-defect care, their quality profiles diverged across domains of completeness, precision, transparency, and contextual fit. Clinicians should therefore recognize both the utility and constraints of such models when consulting them for professional insight.
%U https://tsdp.net/article/investigation-of-large-language-models-capabilities-in-answering-clinical-questions-related-to-peri-a7ssczwr78ab50k