<![CDATA[The use of artificial intelligence in diabetes therapy for dose optimization and safety monitoring of antidiabetic drugs has increased substantially over the past decade. This scoping review was conducted to map the types of AI models applied, to evaluate their impact on glycemic control, and to analyze their contribution to strengthening pharmacovigilance systems. Approaches including machine learning, deep learning, and reinforcement learning have been implemented to model nonlinear dose–response relationships and to identify plateau effects. Adaptive dosing recommendations have been generated using clinical data and continuous glucose monitoring inputs. Improvements in time in range and reductions in HbA1c levels have been reported in comparison with conventional therapeutic approaches. In drug safety monitoring, detection and analysis of adverse drug reactions have been enhanced through the application of natural language processing, Bayesian modeling, and generative AI. Data extraction from electronic health records and individual case safety reports has been performed more efficiently and systematically. Causality assessment processes have been accelerated, leading to improved efficiency in risk evaluation. AI integration in diabetes management has also been implemented through closed-loop systems, real-time glucose prediction, and identification of patients at risk of inappropriate dosing.Several methodological and regulatory challenges remain, including data bias, limited external validation, and concerns regarding algorithmic transparency. The need for real-world validation and strengthened ethical and governance frameworks has been identified to ensure safe and accountable clinical implementation]]>
Copyrights © 2025
