Predicting knee osteoarthritis (KOA) progression typically requires combining diverse data types and expertise from multiple clinical domains. A new framework called OA-MAP addresses this by using multiple AI agents, each handling different aspects of the problem.
The framework is designed to produce evidence-grounded predictions, meaning it can show the reasoning behind each risk estimate. This is a shift from isolated risk scores, which offer limited insight into the factors driving progression.
By integrating multimodal data and multidomain knowledge, OA-MAP aims to support patient monitoring with more interpretable outputs. The approach is described in a preprint on arXiv, and as a single source, it has not yet been independently validated.