摘要
目的通过整合多维度临床指标,基于多因素Logistic回归筛选出卵巢储备功能的影响因素,构建列线图式卵巢储备功能预测模型。方法采用回顾性研究设计,纳入2017年1月至2024年11月在宁夏医科大学总医院生殖医学中心接受辅助生殖技术(ART)助孕的544例女性患者,其中卵巢储备功能减退(DOR)患者203例、卵巢储备功能正常(NOR)患者341例。通过系统收集患者基线特征、血清学指标及超声参数,采用单因素分析筛选潜在预测因子,多因素Logistic回归确定独立影响因素,构建列线图预测模型;模型性能通过Bootstrap重复抽样法重复抽样1000次进行内部验证。结果共纳入训练集人群381例、验证集人群163例。Logistic回归分析确定女方年龄较大、存在盆腔手术史、基础FSH水平升高、抗苗勒管激素(AMH)水平降低、窦卵泡计数(AFC)减少、体质量指数(BMI)升高为DOR的独立影响因素(P<0.05)。将这6个独立影响因素纳入预测模型,当综合得分达240分时DOR患病概率达50%,而当综合得分达290分时DOR患病概率上升至90%。模型验证通过绘制ROC曲线,训练集及验证集的AUC分别为0.837[95%CI(0.794,0.879)]、0.834[95%CI(0.768,0.900)],表示该预测模型区分度较好;校准曲线显示预测概率与实际概率高度一致,表现出稳定的校准性能;临床决策曲线(DCA曲线)表明训练集与验证集中当阈值概率在15%~60%范围内具有稳定的临床净获益。结论该预测模型具有较好的区分度、校准度以及临床实用性。根据综合得分将DOR风险等级划分为低(<240分)、中(240~<290分)、高风险(≥290分),临床医生可根据风险分层为个体化促排卵方案的选择提供量化依据。
Objectives:To identify influence factors for ovarian reserve function and construct a nomogram-based prediction model for ovarian function by means of integrating multi-dimensional clinical indicators and multi-factor logistic regression.Methods:A retrospective study design was employed.The study enrolled 544 female patients who underwent assisted reproductive technology(ART)treatment at the Reproductive Center of The General Hospital of Ningxia Medical University between January 2017 and November 2024.Among these patients,203 suffered from diminished ovarian reserve(DOR),while 341 had normal ovarian reserve(NOR).The baseline characteristics,serological indicators,and ultrasound parameters of all patients were meticulously collected.Univariate analysis was utilized to screen for potential predictive factors.Subsequently,multivariate logistic regression was conducted to identify independent influence factors.Based on the results,a nomogram prediction model was developed.To assess the model’s performance,internal validation was carried out using the Bootstrap resampling method with 1000 repeated samplings.Results:A total of 381 patients were recruited into the training set and 163 into the validation set.Logistic regression analysis identified that older maternal age,history of pelvic surgery,elevated basal follicle-stimulating hormone(FSH)level,decreased anti-Müllerian hormone(AMH)level,reduced antral follicle count(AFC),and increased body mass index(BMI)were independent influencing factors for DOR(P<0.05).These six independent variables were then incorporated into the predictive model.When the composite score reached 240,the incidence risk of DOR was 50%.When the composite score reached 290,the probability of DOR increased to 90%.To validate the model,a receiver operating characteristic(ROC)curve was plotted.The area under the curve(AUC)of the training set and the validation set were 0.837[95%CI(0.794,0.879)]and 0.834[95%CI(0.768,0.900)],respectively,suggesting that the predictive model exhibited excellent discriminatory ability.The calibration curve indicated a high degree of consistency between the predicted probabilities and the observed probabilities,demonstrating stable calibration performance.The clinical decision curve(DCA)analysis revealed that within the threshold probability range of 15%~60%in both the training set and the validation set,there was a stable net clinical benefit.Conclusions:This predictive model has good discrimination,calibration,and clinical practicability.Based on the comprehensive score,the risk levels of DOR are classified as low(<240 points),medium(240~<290 points),and high risk(≥290 points).Clinicians can use the risk stratification to provide quantitative evidence for the selection of individualized ovulation induction plans.
作者
许晓雪
华蔓月
何蕊
杨萌
哈灵侠
刘春莲
XU Xiao-xue;HUA Man-yue;HE Rui;YANG Meng;HA Ling-xia;LIU Chun-lian(Ningxia Medical University,Yinchuan 750004;The General Hospital of Ningxia Medical University,Yinchuan 750004;PKUFH-NINGXIA Women&Children’s Hospital(Maternal&Child Health Hospital of the Autonomous Region),Yinchuan 750004)
出处
《生殖医学杂志》
2025年第12期1655-1664,共10页
Journal of Reproductive Medicine
关键词
卵巢储备功能减退
辅助生殖技术
列线图
预测模型
影响因素
Diminished ovarian reserve
Assisted reproduction technology
Nomogram
Prediction model
Influence factor
