AI-enabled integration of life cycle assessment in building retrofitting: A Comprehensive Review of Methods and Applications

ABSTRACT Background Life cycle assessment (LCA) is widely used to evaluate the environmental impacts of building retrofits, yet conventional methods struggle with dynamic data and complex decision-making. Artificial intelligence (AI) offers tools to enhance LCA through predictive modelling, optimisation, and automated data integration. Aim This review synthesises current applications of AI-enabled LCA in building retrofitting, categorising techniques, application areas, and supporting data ecosystems, while identifying challenges and opportunities. Methodology A systematic review of 52 peer-reviewed studies from the last decade was conducted. In the assessment of the studies, a structured quality assessment framework was employed which depended on the type of validation, quality of the data source used, level of application, and the way of handling uncertainties. Performance metrics of the algorithm (R 2, RMSE, MAE), the level of maturity and the appropriateness in the context of retrofit were also compared. Findings Machine learning methods, including regression, decision trees, and neural networks, have been applied to predict retrofit performance, energy use, and life cycle impacts. Comparative analysis of cross-study showed that ensemble techniques (RF, GB) yielded better accuracy (R 2 or greater than 0.90) than single algorithms but at higher computational cost. Optimisation algorithms, such as genetic algorithms and reinforcement learning, support multi-objective trade-offs between cost, carbon emissions, and comfort. There is however a great variation in the depth of validation with most studies being of a simulation only type instead of practical implementation. Automated data management using BIM, IoT sensors, and smart meters enables real-time updates and more accurate assessments. Applications span component-level retrofits in HVAC, glazing, insulation, and lighting, as well as whole-building lifecycle evaluations. Key challenges include data heterogeneity, limited interoperability between AI and LCA platforms, model generalisation, and the transparency of AI outputs. Conclusion AI enhances the scope, accuracy, and adaptability of LCA in building retrofitting, though further progress is needed in standardisation, transparency, and large-scale validation for industry-wide adoption. Key limitations in the service-life model and uncertainty quantification were established in the reviewed literature.