The sustainable development and preservation of natural resources have highlighted the critical need for the effective maintenance of civil engineering infrastructures. Recent advancements in technology and data digitization enable the acquisition of data from sensors on structures like bridges, tunnels, and energy production facilities. This paper explores “smart” uses of these data to optimize maintenance actions through interdisciplinary approaches, integrating artificial intelligence in civil engineering. Corrosion, a key factor affecting infrastructure health, underscores the need for robust predictive maintenance models. Supervised Machine Learning regression methods, particularly Random Forest (RF) and Artificial Neural Networks (ANNs), are investigated for predicting structural properties based on Non-Destructive Testing (NDT) data. The dataset includes various measurements such as ultrasonic, electromagnetic, and electrical on concrete samples. This study compares the performances of RF and ANN in predicting concrete characteristics, like compressive strength, elastic modulus, porosity, density, and saturation rate. The results show that, while both models exhibit strong predictive capabilities, RF generally outperforms ANN in most metrics. Additionally, SHapley Additive exPlanation (SHAP) provides insights into model decisions, ensuring transparency and interpretability. This research emphasizes the potential of integrating Machine Learning with empirical and mechanical methods to enhance infrastructure maintenance, providing a comprehensive framework for future applications.