Objective To develop a machine learning predictive model to assess the risk of gastrointestinal symptom occurrence in patients after laparoscopic cholecystectomy (LC) and validate its predictive performance. Methods A retrospective cohort of 96 patients who underwent LC from October 2022 to October 2024 was included, and clinical data were collected. Missing values were handled using multiple imputation. Random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost), and logistic regression (LR) models were constructed using R language. The model with the highest performance was selected based on area under the curve (AUC), sensitivity, and specificity, and its performance was validated using calibration curves. Results Within 72 h post-operation, 27 patients (28.13%) experienced gastrointestinal symptoms, while 69 patients (71.87%) were asymptomatic. In the least absolute shrinkage and selection operator (LASSO) regression analysis, five non-zero features were selected, including age, surgery duration, anesthesia duration, gallbladder wall thickness, and preoperative C-reactive protein (CRP). The AUC for the RF, XGBoost, LR, and SVM models were 0.855, 0.893, 0.762, and 0.839, respectively. The XGBoost model showed the best balance between sensitivity (85.71%) and specificity (82.35%), demonstrating the optimal predictive performance. A nomogram was constructed based on this model, providing moderate net benefit in predicting the incidence of gastrointestinal symptoms in postoperative LC patients. The Hosmer-Lemeshow goodness-of-fit test indicated that the predictive model had good calibration ability (χ2＝2.176, P＝0.693). Conclusion The XGBoost-based machine learning nomogram effectively predicts the risk of gastrointestinal symptoms after LC. Integrating preoperative imaging and inflammatory indicators can assist in optimizing surgical strategies and postoperative management.