ACR 2026: Artificial Intelligence Enhances Ultrasound and Clinical Decision Support Processes in Rheumatology

At ACR Convergence 2026, the annual meeting of the American College of Rheumatology, three separate studies were shared on how AI-supported systems can improve diagnosis and education processes in rheumatologic diseases.

Researchers developed a deep learning platform that enables the automated detection of synovitis, power Doppler signals, bone erosions, and osteophytes in ultrasound scans. This system assists in differentiating inflammatory arthritis from degenerative joint changes, providing reproducible scoring metrics. However, experts state that this technology must undergo multi-center validations with different patient groups and devices before being integrated into clinical workflows.

Other studies focused on the necessity of a dedicated curriculum for physicians to understand the limitations of AI models. The training programs emphasize the recognition of algorithmic biases and the importance of independent clinical validation of automated outputs. Additionally, the adherence of widely used generative AI models to medical recommendations was analyzed; researchers found that although these models generate persuasive texts, they can contain a significant amount of erroneous information (hallucinations) and unvalidated models cannot be used as clinical authorities.