Establishing prediction model for retained cesarean scar pregnancy after curettage based on LASSO regression / 中国医学影像技术

ABSTRACT

Objective To establish a risk prediction model,and to observe its value for predicting retained cesarean scar pregnancy(CSP)after ultrasound-guided curettage.Methods Data of 401 CSP patients who received ultrasound-guided curettage were retrospectively analyzed.The patients were randomly divided into training set(n=264)or validation set(n=137)at a ratio of 7:3.According to whether there was retained CSP at the lower segmental scar of uterine after ultrasound-guided curettage,the patients were divided into retained group or non-retained group.The variables with the biggest predictive value for retained CSP after ultrasound-guided curettage were selected with LASSO regression,and the independent risk factors were screened using multivariate logistic regression,and then a nomogram model was established.Results The results of LASSO regression and multivariate logistic regression indicated that embedded depth of gestational sac in cesarean scar more than 1.13 cm,convexity of gestational sac,rich blood supply(Adler degree Ⅱ-Ⅲ),and pre-curettage serum β-human chorionic gonadotropin(HCG)more than 33 063.50 U/L were all independent risk factors for retained CSP after curettage(all P＜0.05).The calibration curve of nomogram predictive model established based on the above indexes was basically consistent with the ideal curve,and the model had good clinical benefits.Conclusion The established nomogram predictive model had good predictive ability for retained CSP after curettage.