A systematic review on sustainable soil stabilization using machine learning and deep learning

Abstract

This study investigates how soil stabilizing techniques are becoming more ecologically sustainable. Due in large part to their detrimental impact on environmental pollution, the construction industry is currently under intense pressure to lessen its reliance on conventional binders, such as cement and lime. As a result, focus has switched to more environmentally friendly substitutes, such as natural fibers and industrial byproducts. These materials frequently operate inconsistently, and engineers are using machine learning and deep learning approaches to better forecast how these materials will behave in practical applications. A systematic bibliometric review of papers published between 2010 and 2025 was conducted in order to comprehend the evolution of research in this field. Using specific keyword combinations like “ground stabilization” and “circular economy,” pertinent literature was found in the PubMed, Google Scholar, ResearchGate, ScienceDirect database. After the chosen studies were arranged into a structured review matrix, VOSviewer was used to examine research trends and pinpoint top academic institutions. The results show a significant rise in the application of data-driven methods, pointing to a definite change in soil engineering toward the integration of cutting-edge computational tools with sustainable materials. This review provides a clear summary of advancements throughout the last fifteen years by following strict academic requirements. It also outlines key research gaps, points to the most promising materials and technologies currently in use, and provides direction for future efforts aimed at creating more durable and sustainable construction practices.