A Dynamic Multimodal Predictive Intelligence System Utilizing Natural Language Analytics, Maritime Tracking, and Macroeconomic Indicators for Global Supply Chain Disruption Management

• Ada John

  ladoke Akintola university of technology

  Author

Global supply chains face unprecedented volatility from cascading disruptions—pandemics, geopolitical conflicts, and climate events—that conventional risk management systems cannot anticipate with sufficient lead time. Existing predictive frameworks remain siloed, relying on either structured economic data or unstructured text signals, but rarely integrating maritime tracking, natural language analytics, and macroeconomic indicators into a unified architecture. This study develops and validates a dynamic multimodal predictive intelligence system (DMPIS) that fuses these three data modalities for supply chain disruption forecasting. Using a design-based research approach with retrospective data spanning January 2022 through December 2025, the system was trained and tested on a global dataset covering 25 countries, 88 product categories, and five major maritime hubs. The Transformer-based multimodal fusion architecture achieved 89.4% prediction accuracy, with a 7-day-ahead precision of 87.2% and mean lead time of 12.3 days—outperforming single-modality baselines by 23.7%. Maritime AIS trajectory anomalies emerged as the strongest leading indicator (weight = 0.41), followed by news sentiment volatility (0.33) and macroeconomic divergence metrics (0.26). The framework provides practitioners with an actionable early-warning capability that transforms fragmented data streams into coherent, forward-looking risk intelligence. The study contributes both a replicable technical architecture and empirical evidence that multimodal fusion significantly enhances disruption prediction in globally distributed supply networks.

• pdf

Copyright (c) 2026 Ada John (Author)

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