Logistics supply chain risks mainly come from multiple factors such as transportation delays, inadequate inventory management, supplier defaults, and fluctuations in market demand. Their unpredictability and potential destructiveness pose significant challenges to the stability and resilience of the supply chain. Traditional risk management methods are usually based on empirical rules and statistical models. Although they have certain practical value in specific scenarios, they show obvious limitations when dealing with multidimensional data and complex non-linear relationships in the supply chain. Therefore, this study proposes a supply chain risk prediction method based on the XGBoost model. Combining the powerful non-linear processing capabilities of machine learning technology, it makes full use of key feature data such as transportation delays, inventory turnover rates and supplier default rates to achieve accurate prediction of supply chain risks. Experimental results show that the XGBoost model outperforms traditional methods in terms of multiple performance indicators, including root mean square error (RMSE), mean absolute error (MAE), and area under the receiver operating characteristic curve (AUC). Among them, RMSE and MAE are 0.28 and 0.21, respectively, which are significantly lower than those of the linear regression and random forest models; AUC reaches 0.92, indicating that the model has excellent performance in the risk classification task. At the same time, the study found that transport delays are the most important factor affecting supply chain risk, while inventory turnover rate and supplier default rate also play a key role. This analysis not only verifies the predictive ability of the model, but also provides a clear direction for optimisation for supply chain managers. This study provides a data-driven decision support tool for supply chain risk management, and its results are of great significance for improving supply chain resilience, optimising resource allocation, and coping with uncertainty in complex market environments.