A Multi-Modal Decision-Support Framework for Real-Time Urban Infrastructure Vulnerability Assessment and Emergency Resource Allocation During Extreme Weather Events

• Ada John

  ladoke Akintola university of technology

  Author

Extreme weather events increasingly threaten urban infrastructure systems, yet emergency managers lack integrated tools that can simultaneously assess vulnerability and optimize resource allocation in real time. Existing approaches remain fragmented: hazard prediction models operate independently of resource optimization algorithms, and most systems rely on unimodal data streams that fail to capture the multidimensional nature of urban crises. This study develops and validates a multi-modal decision-support framework that integrates meteorological data, infrastructure sensor networks, and geospatial indicators to predict vulnerability and allocate emergency resources dynamically during flood events. The framework employs a hybrid deep learning architecture combining LSTM networks for temporal forecasting with gradient boosting for vulnerability classification, coupled with a stochastic optimization module for resource allocation. Validation using historical flood data from three urban regions demonstrates that the framework achieves 89.4% accuracy in identifying high-vulnerability infrastructure nodes within a six-hour lead time. The integrated approach reduces resource allocation latency by 42% compared to static allocation methods and improves coverage equity by 28%. These findings establish that multi-modal integration significantly enhances both predictive accuracy and operational efficiency in emergency management. The framework provides a replicable model for municipalities seeking to strengthen urban resilience through data-driven decision support, with implications extending to other climate-sensitive infrastructure systems and emergency contexts.

• pdf

Copyright (c) 2026 Ada John (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.