Exactitud y coherencia de plataformas de inteligencia artificial vs diagnóstico clínico periodontal. Clínica de periodoncia de la UCSG.

Abstract

Introduction: Artificial intelligence (AI) has assumed a significant role in the healthcare field due to its capacity to facilitate information automation and large-scale data processing. In dentistry, its implementation has raised questions regarding how capable these platforms are in periodontal diagnosis, considering that periodontitis represents a highly prevalent public health problem and that one third of cases may be misclassified. Objective: To evaluate the accuracy and consistency of responses generated by artificial intelligence platforms in comparison with clinical periodontal diagnosis. Materials and methods: A quantitative, cross-sectional comparative study was conducted with a sample of 265 clinical records selected through non-probability convenience sampling. The platforms ChatGPT, Gemini, and Character.ai were evaluated by comparing stage, grade, and extent according to the 2018 AAP–EFP classification. Data were analyzed using contingency tables and the Kappa concordance coefficient. Results: ChatGPT and Gemini demonstrated greater accuracy and agreement in identifying the stage and extent of periodontal disease. Grade classification showed poor performance across the three evaluated platforms. Character.ai exhibited lower consistency compared to the other platforms. Conclusions: AI platforms may be considered a complementary tool for the diagnosis of periodontitis; however, they do not replace professional judgment in clinical practice.