Objective:To develop and validate a clinical prediction model for assessing the risk of uterine fibroid occurrence based on serum vitamin D levels and other clinical characteristics.Methods:A retrospective analysis was conducted on 304 pa-tients with uterine leiomyomas and 184 healthy controls.They were divided into a training set(n=192)and a validation set(n=296)according to different time periods of data collection.Five machine learning(ML)predictive models were developed based on serum vitamin D levels and five other clinical features.The area under the curve(AUC)was used to evaluate the per-formance of the models,and external validation was performed to verify the models.Results:All five machine learning models demonstrated good predictive performance(AUC≥0.8).In external validation,the support vector machine(SVM)model ex-hibited superior predictive performance compared to other machine learning algorithms(AUC=0.941,95%CI:0.927～0.969,sensitivity=0.968,specificity=0.801,Brier score=0.148).Conclusion:The combination of serum vitamin D levels with five other clinical features shows promising potential for predicting the occurrence of uterine fibroids.Among the models tested,the support vector machine(SVM)model demonstrated the best predictive performance.