Objective To compare the predictive performance of 3 risk prediction models based on machine learning in predicting the risk of gastrointestinal dysfunction among patients with acute stroke-related dysphagia. Methods Using convenience sampling me-thod, 1,095 hospitalized patients with acute stroke-related dysphagia were selected and randomly divided into a training set and a test set according to the 7∶3 ratio. Three types of risk prediction models were constructed based on machine learning for gastrointestinal dysfunction in the participants, namely, logistic regression, decision tree, and random forest. The accuracy, precision, sensitivity, specificity, F1 score and the area under the receiver operating characteristic curve (AUC) were used to compare the predictive performance of the 3 models. Results The incidence of gastrointestinal dysfunction among the participants was 49.41%. Both the random forest and logistic regression models indicated that higher scores on the National Institutes of Health Stroke Scale (NIHSS) and the Standard Swallowing Function Assessment Scale, and elevated ratios of monocyte-to-lymphocyte, neutrophil-to-lymphocyte, and platelet-to-lymphocyte, decreased albumin levels and stroke location in the brainstem or cerebellum were risk factors of gastrointestinal dysfunction in patients with acute stroke and dysphagia (all P＜0.05). As for the logistic regression mo-del, random forest model, and decision tree model, the accuracy rate was 0.773, 0.803, 0.721, the precision rate was 0.792, 0.805, 0.768, the sensitivity was 0.717, 0.780, 0.604, the specificity was 0.825, 0.825, 0.830, the F1 score was 0.752, 0.792, 0.676, and the AUC was 0.848 (95%CI:0.806-0.890), 0.871 (95%CI:0.833-0.910), and 0.728 (95%CI:0.680-0.776) respectively. Conclusion The random forest predictive model for gastrointestinal dysfunction in patients with acute stroke-related dysphagia performs better than the logistic regression model and decision tree model, so it can serve as a reference for the early clinical identification, prevention and formulation of relevant intervention measures.