Objective To construct and validate a risk prediction model for subthreshold depression (SD) in early pregnancy women undergoing assisted reproductive technology (ART), aiming to provide a reference for early identification of high-risk population. Methods A total of 322 women in early pregnancy who conceived through ART were enrolled and randomly divided into a training set (n=225) and an internal validation set (n=97) at a 7∶3 ratio.Another 100 women were enrolled for external validation.Predictors were screened through univariable analysis, least absolute shrinkage and selection operator (LASSO) regression, and mul-tivariable logistic regression, and a model was constructed.The model performance was assessed via the area under the receiver operating characteristic curve (AUC), Hosmer-Lemeshow test, calibration curves, and decision curve analysis (DCA). Results Totally 131(31.0%) early pregnancy women undergoing ART suffered from SD.Body mass index (BMI) during early pregnancy, occupation type, duration of infertility, reason of infertility, and daily sleep duration were incorporated into the model.The AUCs for the training set, internal validation set, and external validation set were 0.847 (95%CI:0.793~0.902), 0.826 (95%CI:0.743~0.909) and 0.832 (95%CI:0.743~0.922), respectively.The Hosmer-Lemeshow test P-values were 0.399, 0.626 and 0.094, respectively.The calibration curves showed good agreement between predicted and observed probabilities.DCA indicated that the model had favorable clinical applicability. Conclusion Women undergoing ART are at high risk of SD during early pregnancy.The prediction model developed in this study demonstrates good performance and can serve as a reference for clinical identification of high-risk populations for SD.