Machine learning prediction of mechanical properties of concrete: Critical review

•Empirical models for concrete mechanical strength are inaccurate and cannot accommodate new input parameters.•Machine learning models are more accurate, flexible and can be retrained with updated databases.•Advantages and shortcomings of ML models identified model performance is compared.•Recommendations for selecting suitable model are made based on review.•Knowledge gaps and needed future research are identified. Accurate prediction of the mechanical properties of concrete has been a concern since these properties are often required by design codes. The emergence of new concrete mixtures and applications has motivated researchers to pursue reliable models for predicting mechanical strength. Empirical and statistical models, such as linear and nonlinear regression, have been widely used. However, these models require laborious experimental work to develop, and can provide inaccurate results when the relationships between concrete properties and mixture composition and curing conditions are complex. To overcome such drawbacks, several Machine Learning (ML) models have been proposed as an alternative approach for predicting the mechanical strength of concrete. The present study examines ML models for forecasting the mechanical properties of concrete, including artificial neural networks, support vector machine, decision trees, and evolutionary algorithms. The application of each model and its performance are critically discussed and analyzed, thus identifying practical recommendations, current knowledge gaps, and needed future research.