<p>Trauma-induced coagulopathy (TIC) has a high incidence in patients with severe trauma. Patients who develop TIC usually have a poor prognosis, characterised by increased organ dysfunction, susceptibility to sepsis, and high mortality. Nonetheless, there are still few studies specifically focusing on postoperative TIC in severely traumatic patients. Therefore, the aim of this study was to construct a machine learning model for early identification of people at high risk of postoperative TIC. This retrospective analysis included data of severe trauma patients undergoing surgical treatment from January 2013 to February 2023 across four hospitals in China. Data of one hospital (<i>n</i> = 1204) was used for the development dataset, while other three hospitals contributed to the external validation dataset (<i>n</i> = 863). The study employed various machine learning algorithms, including random forests, logistic regression, gradient boosting decision trees, support vector machines, backpropagation artificial neural networks, extreme gradient boosting, and naïve Bayes. Model performance was estimated on the basis of accuracy, sensitivity, specificity, and area under the curve. In the internal cross-validation dataset, Shapley’s additive interpretation was applied to the model with the largest area under the receiver operating characteristic curve. TIC occurred in 25.4% (306/1204) and 2.9% (25/863) of patients in the developing and external validation set, respectively. Among the models evaluated, the Random Forest model demonstrated the highest performance, achieving an area under the curve of 0.82 for the test cohort and 0.73 for the external validation cohort. The findings suggest that machine learning models can effectively identify severely traumatized patients at a higher risk of postoperative trauma-induced coagulopathy. Utilizing machine learning may enhance clinical decision-making and improve management strategies for postoperative coagulation issues.</p>