Inteligência artificial aplicada à triagem de cefaleias secundárias e à estimativa da idade cronológica por meio de medidas mandibulares

Abstract

Introduction: Artificial intelligence has demonstrated increasing applicability in healthcare, offering tools capable of assisting clinical reasoning and predictive data analysis. In dentistry, its use may contribute to the screening of secondary headaches of orofacial origin and to the analysis of imaging exams related to craniofacial growth and chronological age estimation. Objective: To develop and evaluate artificial intelligence–based applications in dentistry, including a chatbot to assist dentists in screening secondary headaches and machine learning models to estimate chronological age from mandibular measurements. Methods: The dissertation consisted of two studies. In the first, Head-Dent was developed, a chatbot based on natural language processing structured according to the criteria of the International Classification of Headache Disorders (ICHD-3), followed by qualitative validation by specialists and quantitative evaluation in simulated scenarios in a controlled environment. In the second study, data from 385 individuals aged 5 to 18 years were analyzed using supervised machine learning algorithms to estimate chronological age based on mandibular morphometric measurements obtained from imaging exams. Performance was evaluated using statistical metrics and cross-validation. Results: The chatbot achieved maximum performance in a controlled environment in identifying and classifying relevant secondary headaches, in addition to high agreement with established diagnostic criteria and positive validation by specialists. In the age estimation study, mandibular measurements showed a significant association with chronological age, and machine learning models demonstrated good predictive capability, with low mean errors and consistent performance across the different algorithms evaluated. Conclusion: Artificial intelligence demonstrated potential as a support tool for dental practice, both in headache screening and in chronological age estimation through imaging exams. The developed chatbot showed high performance in a controlled environment, while predictive models showed promising results in the analysis of mandibular growth. Future studies in real clinical settings are necessary to expand the validation and applicability of these technologies.