Artificial Intelligence-Driven Flight Disruption Management: A Scoping Review and Critical Evidence Synthesis for Predictive Decision Support in Airline Operations Control Centers

Airline Operations Control Centers (OCCs) manage disruptions in a tightly coupled socio-technical system in which weather, airport and airspace constraints, aircraft availability, crew legality, ground processes, and passenger connections interact. This scoping review and critical evidence synthesis examines how artificial intelligence (AI), machine learning (ML), network modeling, and optimization are being used to predict and manage airline operational disruptions and evaluates the evidence for their readiness for OCC decision support. The review is structured around five questions addressing methods, data and validation, the linkage between prediction and recovery, operational deployment barriers, and priorities for translation into practice. A reproducible search protocol is specified for Scopus, Web of Science, IEEE Xplore, ScienceDirect, SpringerLink, ACM Digital Library, and supplementary Google Scholar discovery; verified study-selection counts are intentionally left as submission-stage placeholders rather than fabricated. Evidence is synthesized across flight-delay prediction, delay propagation, aircraft recovery, crew recovery, passenger recovery, integrated recovery, emerging deep-reinforcement-learning approaches, and human centered decision support. The literature indicates comparatively mature methodological foundations for delay prediction and mathematical recovery, but substantially weaker evidence for end-to-end systems that convert forecasts into feasible, explainable, uncertainty-aware recovery actions in live OCC environments. A central finding is that predictive performance and operational decision value are distinct: high accuracy does not itself demonstrate lower disruption cost, improved recovery, reduced workload, better passenger outcomes, or safer operations. The review therefore identifies the prediction-to-action integration gap as a key research problem and proposes validation priorities centered on temporal and external validation, operational feasibility, calibrated uncertainty, human authority, model-drift monitoring, cybersecurity, and real-world shadow-mode evaluation.