AI-driven predictive analytics for supply chain disruption and resilience in engineering management

Disruptions in the supply chain (SC) are occurring more often and becoming more complex, posing significant challenges for engineering management and operational resilience. Current prediction methods do not adequately reflect the relationships and uncertainty among variables, as well as the causal factors of disruption events. In this study, we present a novel uncertainty-aware hybrid framework, namely, the University of Queensland Creative Arts/Creativity Network (UQ-CRGATNet), which introduces Dynamic Graph Attention Networks (DGAT), a tab transformer network for contextual feature learning, resilience-aware representation, Monte Carlo (MC) dropout, and an ensemble meta-learner of Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost). Moreover, SHapley Additive exPlanations (SHAP) and Non-combinatorial Optimization via Trace Exponential and Augmented lagRangian for Structure learning (NOTEARS) enable feature attribution and causal disruption analytics that are human-understandable and support trustworthy decisions. The proposed framework is a joint model that models graph-structured relationships, heterogeneous tabular information, predictive uncertainty, and resilience indicators within a unified learning architecture. The extensive experiments performed on a real-world supply chain dataset show that they significantly outperform conventional machine learning, ensemble learning, and transformer baselines. UQ-CRGATNet achieves excellent Receiver Operating Characteristic Area Under the Curve (ROC-AUC) and Precision-Recall characteristics, with accuracy, precision, recall, F1-score, and Matthews correlation coefficient (MCC) of 99.87, 99.87, 99.89, 99.88, and 99.73%, respectively. The uncertainty-guided ensemble will increase prediction accuracy, while causal analytics will provide actionable managerial insights into the drivers of disruptions. The results show that the proposed framework provides a strong, interpretable, and scalable solution for proactively predicting supply chain disruptions and supporting resilient engineering management.

Authors

Institutions

Publication Details

Journal
Frontiers in Artificial Intelligence
Published
2026-11-17
DOI
https://doi.org/10.3389/frai.2026.1928679
Primary Topic
Supply Chain Resilience and Risk Management
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

AI-driven predictive analytics for supply chain disruption and resilience in engineering management

Rakibul Islam, Arafat Hossain, MD Amaddus Bepary Asif, Md. Eahia Ansari et al.
3 links
Frontiers in Artificial Intelligence
Supply Chain Resilience and Risk Management
article

AI-driven predictive analytics for supply chain disruption and resilience in engineering management

Rakibul Islam, Arafat Hossain, MD Amaddus Bepary Asif, Md. Eahia Ansari, Md Rafiul Mahmud, Mahathir Mohammad
article en

Abstract

Disruptions in the supply chain (SC) are occurring more often and becoming more complex, posing significant challenges for engineering management and operational resilience. Current prediction methods do not adequately reflect the relationships and uncertainty among variables, as well as the causal factors of disruption events. In this study, we present a novel uncertainty-aware hybrid framework, namely, the University of Queensland Creative Arts/Creativity Network (UQ-CRGATNet), which introduces Dynamic Graph Attention Networks (DGAT), a tab transformer network for contextual feature learning, resilience-aware representation, Monte Carlo (MC) dropout, and an ensemble meta-learner of Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost). Moreover, SHapley Additive exPlanations (SHAP) and Non-combinatorial Optimization via Trace Exponential and Augmented lagRangian for Structure learning (NOTEARS) enable feature attribution and causal disruption analytics that are human-understandable and support trustworthy decisions. The proposed framework is a joint model that models graph-structured relationships, heterogeneous tabular information, predictive uncertainty, and resilience indicators within a unified learning architecture. The extensive experiments performed on a real-world supply chain dataset show that they significantly outperform conventional machine learning, ensemble learning, and transformer baselines. UQ-CRGATNet achieves excellent Receiver Operating Characteristic Area Under the Curve (ROC-AUC) and Precision-Recall characteristics, with accuracy, precision, recall, F1-score, and Matthews correlation coefficient (MCC) of 99.87, 99.87, 99.89, 99.88, and 99.73%, respectively. The uncertainty-guided ensemble will increase prediction accuracy, while causal analytics will provide actionable managerial insights into the drivers of disruptions. The results show that the proposed framework provides a strong, interpretable, and scalable solution for proactively predicting supply chain disruptions and supporting resilient engineering management.

Frontiers in Artificial IntelligenceVol. 9
University of Southern California (US), California State University Los Angeles (US), International American University (US)
Industry, innovation and infrastructure
Supply Chain Resilience and Risk Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon