Optimization of management plan with a machine learning model for ovarian torsion cases: operative vs. conservative.

Background: Ovarian torsion is a gynecologic emergency requiring timely management to preserve ovarian function and future fertility. Clinical presentation is heterogeneous and frequently non-specific, making the decision between operative and conservative management challenging, particularly in reproductive-aged patients. Objective: Develop and evaluate a machine learning-based prognostic model to support individualized management decisions in ovarian torsion. Methods: We conducted a retrospective cohort study of 219 patients diagnosed with ovarian torsion. Demographic characteristics, reproductive and surgical history, comorbidities, presenting symptoms, laboratory findings, and imaging features were analyzed. Multiple supervised machine learning models were trained and compared, including Logistic Regression, Decision Tree, Random Forest, Gradient Boosting, and Neural Network classifiers. Class imbalance was addressed using class weighting and oversampling techniques (ROSE, SMOTE, SMOTE-ENN). Model performance was evaluated using area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Principal component analysis (PCA) and unsupervised clustering were performed to identify latent clinical domains and phenotypic subgroups. Results: The class-weighted Decision Tree demonstrated the most balanced and clinically interpretable performance (AUC = 0.76; sensitivity = 0.75; specificity = 0.73). Key predictors included doppler findings, BMI, blood group (ABO), ethnicity, major and associated clinical symptoms, the presence of pelvic mass, menopause, history of OCP, infertility and previous surgery. PCA identified eight clinical domains, and clustering revealed two phenotypically distinct patient profiles, underscoring the heterogeneity of ovarian torsion presentations. Conclusion: Machine learning models can support individualized risk stratification in ovarian torsion. A transparent, non-overfitted Decision Tree model offers a clinically interpretable framework that may assist fertility-preserving management decisions. Prospective multicenter validation is required before clinical implementation.