A method, system, device, and medium for predicting urinary incontinence in women after hysterectomy

Abstract

The invention discloses a prediction method, system, device, and medium for urinary incontinence after hysterectomy in women, belonging to the field of artificial intelligence technologies. The method includes: obtaining baseline data of a target population; performing feature selection on the baseline data to obtain a key feature set. The feature-selection process adopts filter-based, wrapper-based, and embedded methods. The key feature set includes preoperative urinary incontinence, physical activity, waist circumference, postoperative constipation, a history of hypertension, and a history of delivering a macrosomic infant. An XGBoost-based prediction model for urinary incontinence after hysterectomy is then constructed. The feature set is used to train the prediction model, and the trained model is used to predict the probability of urinary incontinence in a subject undergoing hysterectomy. The invention can systematically analyze risk factors for urinary incontinence after hysterectomy for benign gynecological conditions and construct a risk-prediction model, thereby improving prediction accuracy.