Predictive Analytics and AI-Driven Risk Mitigation Architectures for Enhancing Resiliency in Global Multi-Tiered Supply Chain Logistics Projects

• Abbas Ahsun

  Texas University

  Author

Global multi-tiered supply chain logistics projects face unprecedented disruptions from geopolitical instability, natural disasters, and demand volatility, yet traditional risk management approaches remain reactive and siloed. This research addresses the critical gap in predictive frameworks that can anticipate disruptions across complex supplier networks while recommending proactive mitigation strategies. The study employed a quantitative design-based research methodology, analyzing five years of supply chain operational data (2021–2026) from 247 logistics projects spanning 18 countries, encompassing 1,843 suppliers across four tiers. A hybrid AI architecture integrating Gradient Boosting Machines with Long Short-Term Memory networks was developed and validated against historical disruption events. The proposed framework achieved 89.4% prediction accuracy for tier-1 supplier disruptions with a 14-day lead time, significantly outperforming traditional regression models (72.1%, p<0.001) and static risk matrices (68.3%). Feature importance analysis identified supplier financial health (31.2%), geopolitical risk index (24.7%), and inventory buffer ratios (18.5%) as the most critical predictors. The framework demonstrated 87.6% mitigation effectiveness when deployed prospectively, reducing average disruption recovery time from 23.4 to 6.2 days. This research contributes a validated, replicable architecture for AI-driven supply chain risk mitigation, offering practitioners a data-driven decision support system that transforms reactive crisis management into proactive resilience building. The findings have significant implications for supply chain managers, logistics policymakers, and future research on multi-tier visibility systems.

• PDF

Copyright (c) 2026 Abbas Ahsun (Author)

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