Decision Tree Model Aids Ovarian Torsion Management Decisions (AUC 0.76)

A retrospective decision tree model using clinical/imaging features predicted operative vs. conservative management in ovarian torsion with AUC 0.76, sensitivity 75%, specificity 73% (n=219). Key predictors: Doppler, BMI, pelvic mass. Needs prospective validation.

• Retrospective cohort of 219 patients; class-weighted Decision Tree outperformed other ML models.
• Principal component analysis revealed 8 clinical domains, and unsupervised clustering identified 2 distinct patient phenotypes.
• Limitation: single-center retrospective design without external validation.

RadPigeon summaries are original and for information only. They are not clinical advice.