Journal Article

AI‐Enabled Collaborative Decision‐Making Mechanisms and Efficiency Improvement Paths for Industrial Supply Chains

Wang, Liqin

2026

Overview

Industrial supply chains continue to face inefficiencies, fragmented decision‐making, and disruptions that weaken resilience and competitiveness. Traditional mechanisms often struggle to balance cost minimization, timely delivery, and adaptability under uncertainty, creating a need for advanced solutions. However, existing AI approaches address these challenges in isolation—predictive, prescriptive, and collaborative mechanisms are rarely unified into a single adaptive framework, leaving a critical gap in achieving end‐to‐end supply chain intelligence. This paper proposes an integrated AI‐enabled framework that combines predictive modeling, prescriptive modeling, and collaborative decision‐making to enhance efficiency in industrial supply chains. The predictive layer employs Long Short‐Term Memory (LSTM) networks for demand forecasting and equipment failure prediction, capturing temporal dependencies and irregular demand shocks with high accuracy. The proposed LSTM model demonstrated superior predictive accuracy, achieving an MAE of 0.01755, an MAPE of 0.0283, RMSE of 0.0235, and an MSE of 0.00006, values significantly lower than those reported by existing methods, confirming its capability to capture complex temporal patterns with high fidelity. The prescriptive layer applies Reinforcement Learning (RL) to dynamically optimize routing, scheduling, and inventory strategies, ensuring cost reduction and lead time improvement under varying operational scenarios. Collaborative decision‐making is achieved through decision fusion and rule‐based heuristics, which integrate outputs from predictive and prescriptive models while embedding domain expertise to ensure realistic and implementable strategies. Efficiency improvement paths are evaluated using key performance indicators, including cost reduction, lead time optimization, inventory turnover, service level enhancement, and sustainability metrics, thereby demonstrating measurable operational gains. The novelty of this research lies in its holistic integration of predictive, prescriptive, and collaborative layers into a unified AI‐enabled framework, moving beyond isolated optimization approaches to establish adaptive ecosystems capable of continuous learning and improvement. Industrial supply chains continue to face inefficiencies, fragmented decision‐making, and disruptions that weaken resilience and competitiveness. Traditional mechanisms often struggle to balance cost minimization, timely delivery, and adaptability under uncertainty, creating a need for advanced solutions. This paper proposes an integrated AI‐enabled framework that combines predictive modeling, prescriptive modeling, and collaborative decision‐making to enhance efficiency in industrial supply chains. The predictive layer employs Long Short‐Term Memory (LSTM) networks for demand forecasting and equipment failure prediction, capturing temporal dependencies and irregular demand shocks with high accuracy. The proposed LSTM model demonstrated superior predictive accuracy, achieving MAE of 0.01755, MAPE of 0.0283, RMSE of 0.0235, and MSE of 0.00006, values significantly lower than those reported by existing methods, confirming its capability to capture complex temporal patterns with high fidelity. The prescriptive layer applies Reinforcement Learning (RL) to dynamically optimize routing, scheduling, and inventory strategies, ensuring cost reduction and lead time improvement under varying operational scenarios. Collaborative decision‐making is achieved through decision fusion and rule‐based heuristics, which integrate outputs from predictive and prescriptive models while embedding domain expertise to ensure realistic and implementable strategies. Efficiency improvement paths are evaluated using key performance indicators, including cost reduction, lead time optimization, inventory turnover, service level enhancement, and sustainability metrics, thereby demonstrating measurable operational gains. The novelty of this research lies in its holistic integration of predictive, prescriptive, and collaborative layers into a unified AI‐driven framework, moving beyond isolated optimization approaches to establish adaptive ecosystems capable of continuous learning and improvement.

• Publisher: John Wiley & Sons, Inc,Wiley
• Subject: Accuracy / artificial intelligence / Collaboration / collaborative decision‐making / Cost reduction / Decision making / Efficiency / efficiency optimization / Lead time / Long short-term memory / LSTM forecasting / Optimization / Prediction models / reinforcement learning / Suppliers / supply chain management / Supply chains