This chapter introduces two distinct approaches for RUL prediction, addressing the challenges of small sample data and multi-sample data in subsea equipment maintenance planning. For small sample data, a method is proposed that integrates data and models using intermittent monitoring data to establish a health index (HI) for subsea equipment. By expanding the HI dataset and applying life analysis on the Weibull based distribution, the method converts the HI into reliability metrics, overcoming data scarcity to predict the RUL of subsea valves. This approach provides a probabilistic framework for RUL prediction under limited data conditions, enhancing model fitting and prediction accuracy. For multi-sample data, a hybrid model-data-driven method is introduced, combining Kalman Filter (KF) with DBN. The DBN-KF fusion improves the state estimation of the system by reducing estimation and observation errors while accounting for the uncertainties in degradation and environmental parameters. This method enables more accurate RUL prediction by integrating these uncertainties into the model, which is then applied to subsea Christmas tree valves. Both methods are demonstrated through case studies, offering tailored solutions for different data availability scenarios, enhancing the reliability of RUL prediction and supporting maintenance decision-making.