Sustainable Selection of Recycled and Low-Carbon Building Materials: A systematic Review

Achieving sustainable development goals (SDG) and reducing the construction industry’s environmental impacts depend on the use of low-carbon and recycled building materials. Machine learning (ML) has become a promising decision support tool for assessing qualities, sustainability criteria, and environmental performance as the amount of available material choices keeps increasing. This study systematically reviews ML application in selecting recycled and low-carbon building materials and contribution to sustainable construction. Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) outline, a comprehensive search was conducted through Scopus database involving 1362 published articles between 2016 and 2026 (July) and 226 studies were used for analysis after screening. The result shows that compressive strength performance are predicted using Artificial Neural Networks (ANN), and Support Vector Machines (SVM) between 2018 and 2018. Expansion to other models occurred between 2019 and 2022 optimizing recycled aggregates to enhance sustainable construction using Random Forest (RF) and XGBoost. Transitioning towards ML models such as explainable AI (XAI), sensitivity analysis, and multi-objective optimization emerged for making sustainable material selection rather than just embodied carbon prediction (EC) and lifecycle assessment (LCA). ML was identified through the keyword co-occurrence map as the research central theme associated with compressive strength, recycled aggregate, supplementary cementitious materials, optimization, LCA, and recycling. This implies that current research is focused on predicting properties of materials and recycled concrete mixtures optimization. However, only a few numbers of studies use ML as a thorough framework for making decisions when choosing low-carbon materials based on a variety of sustainability factors, such as EC, cost, and durability. Study concludes that sensitivity analysis, multi-objective optimization, and XAI are emerging area that are underrepresented. Future research should focus on transparent, interpretable, and multi-objective ML frameworks that support sustainable material selection, thereby supporting the growth of low-carbon and sustainable construction practices.