Conhecimento, percepção e confiabilidade de cirurgiões dentistas em relação ao diagnóstico realizado através da inteligência artificial

Abstract

Introduction: Health treatments begin with the correct diagnosis, based on clinical signs and symptoms and complementary exams. In this context, artificial intelligence (AI) is showing potential to reduce the variability resulting from the examiner, thus improving the diagnosis obtained with imaging exams. Objective: The aim of this study was to evaluate the knowledge, confidence, and attitude of Brazilian dentists regarding the diagnosis made through AI. Materials and methods: A cross-sectional observational study was carried out by sending out electronic questionnaires. Dentists registered in Brazil were invited to participate. Recruitment was carried out through invitation on social networks (Facebook, Whatsapp and Instagram), allowing greater heterogeneity of participants. The questionnaires, developed in 'Google Forms', comprised demographic questions, professional qualification, and statements regarding knowledge, usefulness, and confidence about AI. After descriptive evaluation of the data, the variable level of knowledge about AI was compared with the other variables using the chi-square test. Results: A total of 635 questionnaires were answered. The majority of dentists were female (64.5%), with a mean age of 40 years (23 to 81 years) and with some additional education. Most respondents believe that AI is being used in everyday life (89.5%) and in Dentistry (87.9%), but 69,3 % of the participants answered that they have little or very little knowledge about AI and 84,6% stated that they had never attended lectures about AI. Most participants agreed with the use of AI to make diagnosis more reliable (89.7%), especially when used together with the diagnosis performed by the clinician (94.5%). However, 76.3% of the participants answered that in case of disagreement between the diagnosis of the professional and the AI, they would not accept the AI diagnosis as definitive. Higher level of AI knowledge showed association with willingness to incorporate AI in Dentistry (P = 0.002), accepting AI diagnosis as definitive in case of disagreement with the professional’s diagnosis (P = 0.001), unavailability of digitized data (P = 0.004), and data protection failure (P = 0.007); less reported knowledge was related to increasing expenses (P < 0.001), and reduction in the number of jobs (P = 0.004). Conclusion: Most dentists believe AI is a tool to make diagnosis even more reliable if used in conjunction with the professional's diagnosis. Professionals who have more knowledge about AI are more willing to incorporate this technology into their practice and are more likely to accept the AI diagnosis as definitive in case of disagreement with the professional's diagnosis. Thus, the present study raises the concern in order to provide more knowledge about AI to Brazilian dentists.