Predictive Process Monitoring (PPM) focuses on forecasting the progression of partially executed process instances using historical event logs, addressing tasks such as predicting outcomes, next activities, or time remaining until completion. While advancements in predictive modeling, such as Deep Learning and Ensemble Learning, have improved accuracy, they often neglect explainability, a crucial factor for adoption in practice. Recent efforts to incorporate explainable AI (XAI) into PPM have primarily relied on generic XAI techniques, which lack adaptation to PPM’s temporal and structural constraints. My work explores advanced, domain-specific XAI methods to generate compliant and interpretable explanations, integrating prior knowledge to enhance their relevance. Additionally, in our work, we investigate the potential of using these explanations to refine predictive models, aiming to bridge the gap between performance and interpretability in PPM.