The increase in urban population, unbalanced physical development, and the growth of intra-urban traffic volume have raised the necessity of identifying and predicting urban high-risk areas as one of the most important requirements for safety management and reducing urban vulnerability. The aim of the present study is to present an intelligent model based on geographic information systems to identify and predict urban high-risk areas by integrating spatial, environmental, and social data with machine learning algorithms. In this study, a set of variables affecting the formation of high-risk areas, including population density, road network, intensity and type of land use, distance from high-traffic centers, access to urban services, geometric characteristics of roads, records of dangerous events, and other physical and environmental indicators, are collected, standardized, and analyzed in a GIS environment. Then, using machine learning algorithms, including Random Forest, Support Vector Machine, and XGBoost, the spatial pattern of high-risk areas is extracted and the probability of danger in different urban areas is predicted. In order to increase the accuracy of the model, the relative importance of the variables will be evaluated using model interpretability methods and the performance of the algorithms will be compared based on indicators such as Accuracy, Precision, Recall, F1-Score and AUC-ROC. Finally, the model output will be presented in the form of a risk zoning map in a GIS environment to enable the identification of high-risk areas and the prioritization of management and infrastructure interventions.