Identifying vital nodes in complex networks is critical in various research areas, including social network analysis, epidemiology, and physics. Centrality measures are commonly used and combined for this purpose. However, vital node identification is often hindered due to privacy restrictions, particularly in networks built from sensitive data like Bluetooth-based contact networks. This study introduces a machine learning-based approach that leverages the outputs of vital node identification algorithms. Our approach demonstrates that, even when trained on just 20% of the data, our proposed models can significantly outperform state-of-the-art methods, particularly in scenarios where network information is severely limited. This research advances the understanding of privacy-centric methods in complex network analysis and shows how machine learning can enhance vital node identification under privacy-preserving conditions.