AI-Driven Predictive Supply Chain Resilience Framework for U.S. Manufacturing Systems: An Empirical Analysis

This study develops and empirically evaluates an AI-driven predictive supply chain resilience framework tailored for U.S. manufacturing systems, addressing the increasing vulnerability of traditional supply chains to global disruptions. The research is grounded in the transition from reactive, Just-in-Time models to proactive, data-driven resilience strategies aligned with Industry 4.0 principles. A comprehensive manufacturing and supply chain dataset was constructed by integrating operational variables, logistics performance metrics, and external disruption indicators. Advanced machine learning techniques, specifically the Random Forest algorithm, were applied to predict supply chain disruptions, their severity, and recovery timelines. The model demonstrated high predictive performance, achieving an overall accuracy of 91.8%, with precision of 89.6%, recall of 92.3%, and an F1-score of 90.9%. The area under the ROC curve (AUC) reached 0.94, confirming strong classification capability in identifying disruption events. The results further revealed significant operational improvements following the implementation of the predictive framework. Disruption response time was reduced by approximately 35%, while inventory optimization led to a 22% decrease in stockouts and a 15% reduction in excess inventory costs. Logistics efficiency improved, with transportation delays reduced by 18% and service levels increasing to an average of 96%. Additionally, the framework contributed to an estimated cost savings of 12–18% across supply chain operations. These findings highlight the effectiveness of integrating predictive analytics with real-time data for enhancing supply chain visibility and responsiveness. Overall, this study demonstrates that AI-driven predictive models significantly outperform traditional reactive approaches, offering a scalable and data-driven solution for disruption management. The proposed framework provides both theoretical and practical contributions by bridging the gap in real-time predictive resilience modeling and supporting strategic decision-making in complex manufacturing environments.