Machine Learning (ML) interpretability techniques have played a vital role in advancing Predictive Process Monitoring (PPM). This work presents the latest version of the VisInter4PPM framework, which was initially proposed in our previous research as a business-oriented tool to visually support interpretability in PPM. We also present findings from user experiments that tested the interpretability of the framework using both synthetic and real event logs.The framework's utility was validated across two design cycles through expert evaluations, demonstrating its effectiveness in providing interpretable and practical predictions for business process analysts. The evaluations confirmed that VisInter4PPM successfully addressed the needs of business experts by integrating interpretability into familiar process models. This integration not only facilitated but enhanced decision-making in complex business contexts, underscoring the essential role of ML in modern Business Process Management (BPM). The findings suggest that ML in BPM is fundamentally about augmenting human decision-making in the face of growing complexities. This research offers both a methodological framework and empirical evidence, advancing ML transparency in BPM and serving as a key resource for practitioners navigating ML-driven changes in business processes.