AI-Augmented Risk Analytics for Asymmetric Digital Service Supply Chains: Predicting Capacity Shortage and Cyber Exposure
Main article
Abstract
Digital service supply chains increasingly rely on cloud infrastructure, platform software, data interfaces, and external service providers to serve both consumer and enterprise markets. These systems face a compound risk problem: demand fluctuates across channels, capacity commitments are negotiated under asymmetric information, and cybersecurity events can interrupt service availability while they damage trust. This article develops an AI-augmented risk analytics framework for asymmetric digital service supply chains, with a focus on predicting capacity shortage and cyber exposure in dual-channel software-as-a-service operations. A simulation-calibrated data design is used to generate 24,000 firm-period observations reflecting B2C demand volatility, B2B contract intensity, infrastructure latency, information asymmetry, patch delay, failed authentication signals, service capacity, and risk mitigation investment. Five predictive configurations are compared: logistic regression, random forest, gradient boosting, LSTM sequence learning, and a hybrid ensemble that combines structured tabular learning with time-series signals. The hybrid ensemble achieves the strongest performance, with AUC values of 0.893 for shortage prediction and 0.884 for cyber exposure prediction. Sensitivity analysis shows that asymmetric information magnifies capacity shortage risk more strongly than cyber exposure risk, while cyber exposure is most responsive to patch age, failed login intensity, and shared infrastructure dependency. The study contributes a practical analytics architecture, a model comparison benchmark, and managerial guidance for risk governance in digital service supply chains.
