Interpretable artificial intelligence for advancing slope stability assessment techniques with Technosols

AbstractSlope stability is a critical factor in ensuring the safety and longevity of infrastructure, especially in areas prone to landslides and soil erosion. Traditional methods of slope stability assessment, while widely used, often struggle to provide accurate results when applied to Technosols—soils modified by human activities and composed of waste materials. This study proposes a novel approach that combines artificial intelligence techniques to improve the precision of slope stability predictions in these complex soil types. The method utilizes a model based on artificial neural networks, trained on a large dataset of slope stability factors. Unlike conventional techniques, the proposed model integrates multiple environmental and material properties to provide a more accurate assessment compared to other models. The model's performance is demonstrated by R2 values of .999975 for the test datasets, which is significantly better than similar work by statistical analysis. Moreover, by incorporating Shapley Additive Explanations (SHAP), we provide a clear understanding of the impact of various soil parameters on slope stability. The findings suggest that the proposed machine learning‐based method offers a more reliable tool for slope stability evaluation in Technosols, making it a valuable addition to the field.