Abstract: As one of the structural materials for nuclear fuel element fabrication, sponge zirconium reacts with environmental hydrogen, leading to material hydrogen embrittlement. This phenomenon significantly jeopardizes the operational safety and structural reliability of nuclear reactors. Pressure-Composition-Temperature (PCT) Isotherms provide critical insights for regulating the thermodynamic and kinetic behaviors of hydrogen absorption in sponge zirconium. Three predictive models for the PCT isotherms of sponge zirconium were established in this study based on experimentally measured data and data augmentation procedures: polynomial model, support vector regression (SVR) model, and artificial neural network (ANN) model. Compared with the traditional polynomial model, superior prediction accuracy was demonstrated by both the ANN and SVR models, showing MAE reductions of 73.14%% and 63.53% on the test set, respectively. Among the developed models, the ANN approach was demonstrated to exhibit optimal performance in predicting PCT curves under unknown temperature conditions, showing superior generalization capability with a test set R² value exceeding 0.98. An effective approach for precise PCT curve prediction was established through this study for metal-hydrogen systems.