Sustainable and AI-Enhanced Inventory Models under Stochastic Demand and Product Deterioration: A Comprehensive Review

Inventory management in modern supply chains faces growing complexity due to stochastic demand, product deterioration, sustainability requirements, and the need for rapid decision-making. While classical models such as the Economic Order Quantity (EOQ) and continuous review policies have provided foundational insights, they fall short in addressing the combined challenges of demand uncertainty, time-dependent deterioration, and environmental constraints. Recent advancements integrate sustainability considerations, such as carbon emission reduction and energy efficiency, into inventory frameworks, while artificial intelligence (AI) techniques particularly Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRU), and metaheuristic algorithms enhance forecasting accuracy and optimization performance. This review, conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) method, systematically identified, screened, and evaluated relevant literature, resulting in the inclusion of 25 peer-reviewed papers that address economic, environmental, and technological dimensions of inventory modelling. Key findings reveal limited research integrating all these factors into a unified framework, insufficient real-time AI-driven decision-making applications, and an overreliance on carbon emissions as the sole sustainability metric. Additional gaps include the underexplored interaction between financial incentives and preservation technologies, and a lack of industry-specific empirical studies in highly regulated sectors. Addressing these gaps can lead to next-generation inventory systems that are economically viable, environmentally responsible, and technologically adaptive.