Machine Learning Applications for Evaluating Fire Performance of Steel‐framed Buildings: A Review

Machine Learning (ML) is increasingly applied to evaluate the fire performance of steel‐framed buildings, from predicting material behaviour at elevated temperatures to estimating member resistance and system‐level collapse. Recent advances include interpretable models, physics‐informed methods, and emerging surrogate strategies that embed engineering knowledge into data‐driven learning. This review summarizes existing methods and datasets, highlights challenges related to generalizability and uncertainty, and identifies future research needs such as standardized benchmarks, rigorous validation, and integration into design codes. The advantages and limitations of current approaches are discussed, and recommendations are provided to advance ML‐based fire safety design in steel‐frame structures.