LAUSR

Osteoarthritis (OA) is the most prevalent form of arthritis, commonly affecting the knee joint and characterized by the progressive degeneration of articular cartilage (AC). Among the various contributing factors, repetitive cyclic loading plays a significant role in accelerating this deterioration. Knee osteoarthritis (KOA), in particular, represents a big data challenge due to the complexity, heterogeneity, and large volume of data required for its analysis and prediction. To address this, we employed a validated mathematical model capable of predicting the number of remaining mechanical cycles that the AC can endure during daily walking before showing signs of degradation. This model facilitated the generation of a wide range of simulations and scenarios that incorporated diverse individual profiles, including age, height, weight, gender, walking duration, and accumulated cartilage damage, along with their corresponding remaining mechanical cycles. The objectives of this study are twofold: (i) to use the degradation model’s outputs to train, validate, and test four machine learning (ML) models, and (ii) to compare the best-performing ML model with a Long Short-Term Memory (LSTM) neural network. Among the models tested, the Support Vector Regressor (SVR) demonstrated superior predictive performance, achieving an R2 of 0.95, a Root Mean Squared Error (RMSE) of 0.13, and a Mean Absolute Percentage Error (MAPE) of 2.5%. These findings provide a solid foundation for accurately predicting knee cartilage damage and guiding the prescription of personalized treatment strategies aimed at delaying or preventing the onset of KOA.