Objective To develop and internally validate a risk prediction model for postpartum stress urinary incontinence (SUI) in women using machine learning, aiming to provide evidence for early prevention. Methods This retrospective study included 516 postpartum women who underwent pelvic floor screening at 42 days after delivery at Ningbo University Women and Children's Hospital from October 2022 to June 2023 (a case group of 82 cases and a control group of 434 cases). The influencing factors were collected, including age, delivery mode, parity, electromyographic(EMG) data of pelvic floor and pelvic floor ultrasound data. Five machine learning algorithms such as random forest (RF) and support vector machine (SVM) were used to build predictive models. Model performance was evaluated using 10-fold cross-validation. Results The overall incidence of postpartum SUI was 15.89%. Among all models, the random forest model demonstrated the best performance, with an F1 score of 1.000 and AUC of 1.000 in the training set, and an F1 score of 0.776 and AUC of 0.837 in the validation set. The key influencing factors included pelvic floor EMG parameters (e.g., intense contraction and endurance contraction phases), age, parity, perineal injury, and BMI. Pelvic floor EMG parameters contributed the most to classification performance. Conclusions The random forest model shows the best predictive ability. Pelvic floor EMG parameters are strongly associated with postpartum SUI. This model offers a valuable tool for clinicians to accurately assess individual risk and guide early rehabilitation intervention, such as Kegel exercises and biofeedback therapy, which provides new tools for the prevention and treatment of postpartum SUI.