Nomogram for predicting poor outcomes after hysteroscopic endometrial polypectomy in menstrual disorders.

Patients with endometrial polyps often exhibit irregular menstrual symptoms, and even after hysteroscopic removal of endometrial polyps, some patients' menstrual disorders do not show significant improvement. This study retrospectively included 450 patients, with 355 patients showing significant improvement in menstrual disorders and 95 patients showing no significant improvement. Analyze the demographic, clinical characteristics, and surgical related factors of patients in these 2 situations, as well as the differences in hormone levels before and after treatment. Further, support vector machine can be used to screen the influencing factors of poor efficacy and construct a risk model. Perform multivariate regression analysis to examine the impact of the risk model and its interaction on poor efficacy. Construct a nomogram to further analyze the risk of poor therapeutic efficacy. The top 5 important features selected by support vector machine include progesterone changes, luteinizing hormone changes, polyp morphology, polyp type, and polyp location. The AUC (area under curve) of the constructed risk model is 0.701, indicating high predictive ability. In addition to the risk model, age, BMI (body mess index), menstrual duration, endometrial thickness, polyp size, and follicle stimulating hormone changes are also significant independent factors for poor therapeutic efficacy. The interaction indicates that the risk model performs better in patients with longer menstrual periods or larger polyps. Nomogram have high accuracy and clinical application value. The risk model we constructed has a good ability to predict poor therapeutic efficacy, and the nomogram provides a reference for clinical doctors' decision-making.