This study addresses a gap in the adaptive management of urban intermodal hubs operating under high-risk conditions, where conventional planning methods are found inadequate during dynamic disruptions and data scarcity. We present a methodology using crowdsourced data for risk-aware demand forecasting and verify our approach through a case study of the Kharkiv-Pasazhyrskyi Railway Station in Ukraine. Our results show that reliable prediction remains feasible even in conflict environments. The approach combines multiple data sources, including social media activity, historical patterns, and real-time risk indicators, within a multiple linear regression model that shows good predictive performance (MAE = 58, RMSE = 74). While not a complete digital twin implementation, this research establishes the demand-simulation foundation for such systems, confirming that interpretable forecasting can be accomplished with open and synthetic data. The study offers a practical framework for resilient transport management, creating the basis for cost-effective and rapidly deployable adaptive systems that can maintain operational continuity in cities facing emergencies. By measuring risk impacts and enabling proactive response strategies, our work shows potential to support passenger safety and service reliability in conflict-affected urban areas.