Estimation of Time-to-Total Knee Replacement Surgery with Multimodal Modeling

Motivation: Accurate time-to-total knee replacement (TKR) predictions are essential for effective patient management and resource planning.Reliable models for predicting TKR timing are essential to support personalized treatment and enable timely interventions for knee osteoarthritis (KOA) management Goal(s): Develop an AI-based survival analysis model to estimate time-to-TKR using multimodal data, identifying key features associated with KOA progression. Approach: This study used data from the OAI and external datasets, integrating clinical variables, MR images, radiographs, and image assessments in an AI survival analysis framework. Results: The model achieved an accuracy of 73.2% and a C-index of 77.3%, highlighting the effectiveness of multimodal fusion in TKR prediction. Impact: Our model, using AI, survival analysis, and multimodal approaches, enhances TKR decision precision by accurately predicting time-to-TKR within 9 years, supporting personalized osteoarthritis treatment. It enables biomarker exploration and promotes early intervention strategies in knee osteoarthritis.