Artificial intelligence for dental diagnosis and decision support in Latin America

Abstract

Artificial intelligence has been applied in dentistry in Latin America, primarily in imaging-based diagnosis, risk prediction, and clinical decision support. Available studies, largely concentrated in Brazil, employ architectures such as convolutional neural networks, YOLO-based detectors, and machine learning models including XGBoost. Reported performance varies across tasks, with high sensitivity in specific applications but frequent false positives and limited specificity in some approaches, particularly in large language model-based systems. Relevant methodological limitations are also observed, including retrospective designs, small sample sizes, lack of external validation, and heterogeneity in evaluation metrics. The evidence suggests that, despite significant technical advances, widespread clinical implementation in the region remains constrained by validation, infrastructure, and regulatory challenges.

Ähnliche Arbeiten