Supply Chain AnalyticsPub Date : 2026-09-01Epub Date: 2026-08-31DOI: 10.1016/j.sca.2026.100240
Mohammed Zizi, Anas Chafi, Mohammed El Hammoumi
{"title":"Supply chain-informed neural networks: Embedding operational constraints into deep demand forecasting","authors":"Mohammed Zizi, Anas Chafi, Mohammed El Hammoumi","doi":"10.1016/j.sca.2026.100240","DOIUrl":"10.1016/j.sca.2026.100240","url":null,"abstract":"<div><div>Accurate demand forecasting is essential for supply chain planning, yet conventional forecasting models are generally optimized for statistical accuracy without explicitly considering downstream operational requirements. This study proposes Decision-SCINN, a supply-chain-informed neural framework that integrates demand forecasting with operational constraints and decision objectives. Inspired by the broader principle of embedding domain knowledge into neural learning, the framework represents inventory balance through a recursive state transition, incorporates stockout avoidance and service-level targets as soft penalties, and handles capacity and non-negativity through raw-output penalties and a projected decision layer. Decision-SCINN combines forecasting losses with operational penalty terms and a projected decision layer to jointly support predictive accuracy and operational feasibility.</div><div>The framework is evaluated using a public retail demand forecasting benchmark covering the period 2013–2017. Inventory, replenishment, capacity, service-level, and cost variables are generated through a common recursive simulation applied consistently to all evaluated models. Decision-SCINN is compared with LSTM and Transformer forecasting models and with forecast then optimize baselines based on clipping and order up to policies. In the selected main test run, Decision-SCINN achieves an MAE of 6.39, RMSE of 8.33, MAPE of 12.33%, and <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> of 0.933. Under the simulated test policy, it obtains the lowest total operational cost of 7.211 million cost units, maintains a 100% service level, and records no stockout, capacity, non-negativity, or inventory-balance violations in the selected main test run. Although order-up-to baselines also achieve full service and eliminate hard violations, they produce higher operational costs. These findings indicate that integrated constraint-aware learning can provide a more favorable balance between forecasting accuracy, service performance, feasibility, and simulated operational cost than sequential forecast-then-optimize approaches.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"15 ","pages":"Article 100240"},"PeriodicalIF":9.1,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148951025","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supply Chain AnalyticsPub Date : 2026-09-01Epub Date: 2026-08-27DOI: 10.1016/j.sca.2026.100239
Natalia Hartono, Kevin Nathanael, Voicu Ion Sucala, Martino Luis, Elioenai Sitepu
{"title":"A human-centred analytics framework for auditable supply chain risk monitoring and decision support","authors":"Natalia Hartono, Kevin Nathanael, Voicu Ion Sucala, Martino Luis, Elioenai Sitepu","doi":"10.1016/j.sca.2026.100239","DOIUrl":"10.1016/j.sca.2026.100239","url":null,"abstract":"<div><div>Supply chain managers must monitor multi-domain logistics data continuously, detect emergent risks and act faster than manual analysis allows. Large Language Models (LLMs) reason contextually but carry latency and hallucination risk unsuitable for continuous detection. Three gaps persist: AI-enabled supply chain risk monitoring remains predominantly retrospective; no framework implements hybrid LLM-SLM (Small Language Model) deployment in multi-agent architectures or combines rule-based detection with practitioner-facing natural language decision support. This paper addresses these gaps through a human-centred, hierarchical three-tier multi-agent framework that separates detection from explanation: deterministic rules handle rule-violation detection and risk scoring; SLM worker agents, invoked only when rule violations are confirmed, generate domain-specific summaries; and a supervisor LLM synthesises cross-domain findings into a structured risk report. The human operator retains full decision authority at the third tier. Because detection is deterministic and rule-based, it is fast and verifiable, and hallucination is excluded from the detection layer by construction; residual hallucination risk is confined to the generative explanation layer, where human review remains necessary before any operational action. Evaluated on a 32,065-record logistics dataset, the framework screens eight SLM architectures (1–7 billion parameters) and benchmarks four finalists on 100 held-out records. The preferred candidate, Granite 3.1 MoE 3B, reaches 112.2 tokens per second on edge hardware (6 GB VRAM). Compression loss, the omission of items carried by the reference report, accounts for 86.6 per cent of classified divergences. At scale (1000 records), hybrid edge-cloud configuration reduces processing time by 57 per cent and API cost by 39.6 per cent versus fully cloud-based deployment. The evaluation is batch-mode proof of concept; live-stream validation and practitioner user study remain future work. The paper contributes to data analytics through rule-based detection, composite risk scoring, and multi-model benchmarking; and a hybrid edge-cloud AI architecture; and to decision support across supply chain.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"15 ","pages":"Article 100239"},"PeriodicalIF":9.1,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148951024","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supply Chain AnalyticsPub Date : 2026-09-01Epub Date: 2026-09-03DOI: 10.1016/j.sca.2026.100237
Emad Alenany, Abdulrahman Shamsan, Mohammed Ismail Alhussam, Ziang Liu
{"title":"A data-driven constraint learning approach for multi-echelon supply chains","authors":"Emad Alenany, Abdulrahman Shamsan, Mohammed Ismail Alhussam, Ziang Liu","doi":"10.1016/j.sca.2026.100237","DOIUrl":"10.1016/j.sca.2026.100237","url":null,"abstract":"<div><div>Simulation is widely used in supply chain management because it can represent uncertainty and complex network dynamics. Inventory management plays a central role in maintaining high service levels while reducing total supply chain cost. However, optimizing inventory policies is computationally intensive because it typically requires a large number of simulation runs. This paper addresses the problem of identifying satisfactory inventory policies that minimize total supply chain cost in a multi-echelon distribution network subject to target service level and target order frequency constraints. To address this problem, this study proposes a scalable multi-echelon inventory optimization framework that integrates learned constraints into the optimization model, enabling simultaneous consideration of multiple operational requirements without substantially increasing computational effort. First, simulation is used to generate sample points for the multi-echelon distribution network. Next, a machine learning model trained on the generated data is embedded within a mathematical optimization framework. Finally, the <span><math><mrow><mo>(</mo><mi>s</mi><mo>,</mo><mi>S</mi><mo>)</mo></mrow></math></span> inventory policy parameters for each location are determined by solving a mixed-integer programming (MIP) model. Computational experiments on three synthetic divergent distribution networks demonstrate the effectiveness of the proposed framework. Among the evaluated machine learning models, gradient boosting achieves the best performance. Compared with an evolutionary optimization approach, the proposed method reduces computational time by approximately 29%, 81%, and 77% for the two-, three-, and four-echelon networks, respectively. Furthermore, sensitivity analysis provides insights into the impact of order-frequency and service-level targets on total costs.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"15 ","pages":"Article 100237"},"PeriodicalIF":9.1,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148951026","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supply Chain AnalyticsPub Date : 2026-09-01Epub Date: 2026-08-27DOI: 10.1016/j.sca.2026.100241
Ricardo E. Sánchez-Piedrahita, Jairo R. Coronado-Hernández, Germán Herrera-Vidal, Lucas Agudelo
{"title":"An analytical framework for modeling security and drug trafficking supply chains as a complex system","authors":"Ricardo E. Sánchez-Piedrahita, Jairo R. Coronado-Hernández, Germán Herrera-Vidal, Lucas Agudelo","doi":"10.1016/j.sca.2026.100241","DOIUrl":"10.1016/j.sca.2026.100241","url":null,"abstract":"<div><div>Maritime drug trafficking in the Colombian Caribbean exhibits structural properties of a Complex Adaptive System (CAS), characterized by decentralized, evolving, and non-linear interactions. This study establishes a systems-based framework to analyze the adaptive behavior of illicit maritime logistics and extract intelligence for risk-based interdiction. A ten-attribute instrument was designed to quantify CAS dimensions, applied to a sample of 223 experts from military, intelligence, and academic domains. A two-round Delphi technique identified high consensus levels on adaptability, co-evolution, and distributed control as defining system traits. Clustered responses confirmed the presence of self-organization, heterogeneity, and feedback-based learning among network components. Findings were integrated into a decision-support pipeline linking perceptual signatures with spatial interdiction modeling. Critical Success Factors (CSFs), such as dynamic coordination schemes and decentralized resource allocation, were identified as operational enablers. The analysis indicates that linear enforcement doctrines remain insufficient when confronted with adaptive criminal systems. Instead, a data-informed, flexible response architecture emerges as more suitable for countering illicit maritime flows. This framework offers analytical contributions to both security strategy and public policy design. It also advances theoretical understanding by operationalizing CAS principles within the logistics of transnational organized crime.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"15 ","pages":"Article 100241"},"PeriodicalIF":9.1,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148951027","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supply Chain AnalyticsPub Date : 2026-03-01Epub Date: 2025-12-27DOI: 10.1016/j.sca.2025.100191
Sasan Mazaheri , Mahsa Ahmadi , Ali Heidari , Mohammad Hakimi , Mohammad Khalilzadeh
{"title":"An integrated analytics approach to multi-project scheduling and material procurement with coordinated hub location","authors":"Sasan Mazaheri , Mahsa Ahmadi , Ali Heidari , Mohammad Hakimi , Mohammad Khalilzadeh","doi":"10.1016/j.sca.2025.100191","DOIUrl":"10.1016/j.sca.2025.100191","url":null,"abstract":"<div><div>This study presents an integrated framework combining multi-project scheduling, material procurement, and the hub location problem to simultaneously minimize project completion times and overall project and logistics costs. To address the challenge of allocating high-cost renewable resources, we incorporate rental options that balance the trade-off between additional rental expenses and potential project delays. A multi-objective optimization model is developed, integrating the scheduling of multiple projects with coordinated material procurement. To reduce logistics costs and improve delivery efficiency, consolidation hubs are introduced where materials from various suppliers are aggregated before being dispatched to project sites. The model considers the availability of renewable rental resources and storage space capacity while scheduling project activities. Due to the problem's computational complexity, two metaheuristic algorithms NSGA-II (Non-dominated Sorting Genetic Algorithm II) and MOSFS (Multi-objective Stochastic Fractal Search) are employed to obtain near-optimal solutions for large-scale scenarios. The proposed approach is validated through a real-world case study involving a bridge construction project and various benchmark instances of different sizes. Results indicate that while NSGA-II performs better on one performance metric, MOSFS consistently outperforms NSGA-II across most criteria, particularly in large-scale problems. The main contributions of this research include integrating project scheduling, material procurement, and hub location within a single unified framework. The model also incorporates renewable rental resources and realistic, type-specific storage capacity constraints that directly affect material flow and the initiation of project activities.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"13 ","pages":"Article 100191"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076801","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supply Chain AnalyticsPub Date : 2026-03-01Epub Date: 2026-01-26DOI: 10.1016/j.sca.2026.100196
Mahsa Farrokhi , Mauro Gatti , Aram Bahrini
{"title":"An analytics-based structural modeling study of circular practices in sustainable supply chains","authors":"Mahsa Farrokhi , Mauro Gatti , Aram Bahrini","doi":"10.1016/j.sca.2026.100196","DOIUrl":"10.1016/j.sca.2026.100196","url":null,"abstract":"<div><div>The importance of sustainability has significantly increased for many companies in recent years. However, limited studies have explored the integrated impact of circular economy practices, dynamic capabilities, and fields of action on supply chain sustainability. This study addresses this gap by applying circular economy principles within retail‑equipment supply chains to evaluate their impact on the economic, environmental, and social dimensions of sustainability. A comprehensive review of the relevant literature on circular economy, dynamic capabilities, fields of action, and sustainable supply chains facilitated the identification of 51 key indicators. To fulfill the research objectives, a retail-equipping manufacturing company was selected as the focal point of analysis, and data was collected through a structured questionnaire. We analyzed the data via partial least squares structural equation modeling (PLS-SEM) in SmartPLS 3, estimating a reflective–reflective hierarchical component model on n = 131 responses to a 51-item instrument covering circular economy fields, dynamic capabilities, and sustainability outcomes. Results support seven hypotheses: both fields of action (H1–H3) and dynamic capabilities (H4–H6) positively affect economic, environmental, and social performance, and their joint effect is strong (H7: β = 0.73; R<sup>2</sup> = 0.54). The study presents an integrated, empirically validated model that links circular economy dynamic capabilities and fields of action, thereby extending dynamic capabilities theory into the sustainability domain. The findings suggest that circular economy principles not only support environmental protection and cost efficiency but also enhance organizational agility, stakeholder satisfaction, and resilience. While the specific outcomes may vary across industries, this study offers a robust foundation for future research and practical insights to guide strategic decision-making for companies aiming to transition to more sustainable, circular supply chains. The model provides actionable diagnostics to prioritize investments in design, reverse logistics, and capability development aligned with triple-bottom-line goals in manufacturing.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"13 ","pages":"Article 100196"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076800","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supply Chain AnalyticsPub Date : 2026-03-01Epub Date: 2025-12-01DOI: 10.1016/j.sca.2025.100183
Ankur Saurav, Chandra Shekhar, Vijender Yadav
{"title":"A metaheuristic approach to supply chain inventory optimization with rebates, discounts, and emission controls","authors":"Ankur Saurav, Chandra Shekhar, Vijender Yadav","doi":"10.1016/j.sca.2025.100183","DOIUrl":"10.1016/j.sca.2025.100183","url":null,"abstract":"<div><div>This study develops a sustainable production–inventory model that integrates advance booking, rebate incentives, and green investments under carbon cap-and-trade regulations, aiming to optimize profit while ensuring environmental sustainability. The model accounts for time- and price-sensitive demand influenced by discounts and advertising, and organizes the inventory cycle into four phases: advance booking before production, production with ongoing bookings, normal sales, and a deterioration phase supported by rebate strategies. The study introduces two key innovations: a dual-stage advance booking system with associated maintenance costs and deterioration control via preservation technologies. Carbon emissions arising from production, holding, transportation, and deterioration are compared to a regulatory cap, with penalties for exceeding the limit and credit trading for staying within it. Green technology investments further support emission reduction. The model’s nonlinear optimization problem is solved using the Grey Wolf Optimization algorithm to determine optimal pricing, production quantities, investment levels, and cycle times. Numerical results highlight that integrating sustainability measures, smart pricing strategies, and customer incentives enhances profitability while minimizing environmental impact. This model provides valuable insights for firms and policymakers aiming to align operational efficiency with sustainability objectives under regulatory frameworks.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"13 ","pages":"Article 100183"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145694208","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A risk-averse multi-objective analytics framework for green supply chain design under uncertainty","authors":"Shahryar Ghorbani , Javad Nematian , Figen Yıldırım , Selahattin Armağan Vurdu","doi":"10.1016/j.sca.2026.100199","DOIUrl":"10.1016/j.sca.2026.100199","url":null,"abstract":"<div><div>In response to increasing concerns about environmental issues, businesses, and industries face pressure to mitigate their negative environmental impacts. Consequently, firms must reevaluate their operations to align with environmental standards. To address both economic and environmental objectives, industries need to green their supply chains. However, uncertainties in the real world, such as economic instability, add complexity to this greening process. This study proposes a novel risk-averse two-stage stochastic model for green supply chain (GSC) design under uncertainty, integrating Conditional Value at Risk (CVaR) with multi-objective programming. The model uses discrete Fuzzy Random Variables (FRVs) to capture both randomness and fuzziness in cost and emission parameters. To solve the model, we apply possibility theory and Fuzzy Chance-Constrained Programming (FCCP) to derive deterministic equivalents for optimistic, pessimistic, and hybrid decision-making attitudes. Numerical results from a flour supply chain case in Iran show that higher risk aversion increases both cost and CO₂ CVaR, while possibility levels affect outcomes differently across models. The approach provides managers with a flexible tool for balancing economic and environmental goals under uncertainty.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"13 ","pages":"Article 100199"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147394676","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"An analytics-driven hybrid method for multi-item demand forecasting in supply chains","authors":"Md. Limonur Rahman Lingkon , Md. Sanowar Hossain , Ripon K. Chakrabortty","doi":"10.1016/j.sca.2026.100194","DOIUrl":"10.1016/j.sca.2026.100194","url":null,"abstract":"<div><div>This study proposes a new deep learning (DL)–based approach for multi-item demand forecasting in multi-wave distribution networks. In modern merchandising supply systems, traditional forecasting techniques such as moving averages and autoregressive integrated moving average (ARIMA) models are often inadequate for capturing the dynamic nature of sales and operations planning, as they struggle with non-stationary demand, evolving market conditions, and the growing complexity of supply chain networks, resulting in forecasts that are neither sufficiently accurate nor timely. To address these limitations, this study evaluates a hybrid forecasting framework that combines poly-linear regression (PLR) with a transformer-encoder extended long short-term memory (TE-LSTM) architecture to identify latent demand patterns from large and heterogeneous datasets. An empirical analysis compares the proposed PLR–TE–LSTM model with baseline approaches such as standard LSTM across multiple products and distribution locations, demonstrating consistently superior forecasting performance in multi-product, multi-distribution center settings. The study further examines the operational impact of improved forecasting by evaluating alternative inventory review strategies under a fixed-quantity replenishment policy, showing meaningful improvements in forecast accuracy, order fulfillment performance, inventory holding costs, and service levels. The results indicate that the proposed DL framework enhances inventory-related decision-making by reducing excess inventory and stockout risks, thereby improving efficiency and responsiveness in complex distribution networks.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"13 ","pages":"Article 100194"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037467","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A systematic analysis of generative artificial intelligence for supply chain transformation","authors":"Zied Bahroun , Afef Saihi , Rami As’ad , Moayad Tanash","doi":"10.1016/j.sca.2025.100188","DOIUrl":"10.1016/j.sca.2025.100188","url":null,"abstract":"<div><div>Global supply chains face persistent disruptions from geopolitical shocks, sustainability pressures, and volatile demand, creating an increasing need for resilient and transparent operations. Generative Artificial Intelligence (GAI), including Large Language Models (LLMs), Generative Adversarial Networks (GANs), and multimodal generative systems, is emerging as a new decision layer that can generate scenarios, synthetic data, and actionable textual insights rather than only point predictions. This Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-guided systematic review analyzes 98 peer-reviewed studies on GAI applications in Supply Chain Management (SCM) and, to the best of the authors’ knowledge, provides the first combined thematic and Supply Chain Operations Reference (SCOR) model-based mapping of these applications. Publication activity shows a sharp upward trend, with fewer than five papers published before 2021 and 45 published in 2024 alone. Nearly four-fifths of the reported applications focus on the Plan and Enable processes, while the Make and Return processes account for only 4 % and 1 % of the coded functions, respectively. Although LLM- and Generative Pre-trained Transformer (GPT)-based models underpin over 40 % of the implementations, approximately 45 % of the studies do not fully specify their underlying architectures, indicating methodological immaturity. Reported benefits are concentrated in demand forecasting and risk analysis, supplier screening, logistics visibility, and sustainability analytics; however, most evidence remains at the prototype level and rarely reports system-wide Key Performance Indicators (KPIs). The review concludes with a targeted research agenda that emphasizes longitudinal evaluation, hybrid GAI-driven optimization with digital twin architectures, and governance-by-design frameworks to support the responsible and scalable adoption of GAI in supply chains.</div></div>","PeriodicalId":101186,"journal":{"name":"Supply Chain Analytics","volume":"13 ","pages":"Article 100188"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145799697","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}