Abstract: Objective To establish a prediction model of fetomaternal hemorrhage syndrome on machine learning algorithm, so as to assist clinicians in early detection,diagnosis and intervention of FMH. Methods This study was based on the data of pregnant women at 6~42 weeks of gestation who have received antenatal examinations in the Second Xiangya Hospital between June 2019 and December 2020. The recursive feature elimination algorithm was used to select key variables and developed eight kinds of machine learning algorithms and traditional regression model,and the model with the best performance was selected. Besides,the ten-fold cross-validation method was used to verify the model.Results The AUC of XGBoost model is 0.827,the AUC of test set is 0.808,and the accuracy is 0.76. The performance of XGBoost model is significantly better than that of traditional logistic regression model with AUC of 0.681 and other seven machine learning models.Conclusion In this study,a prediction model of FMH based on XGBoost algorithm was successfully constructed and its prediction performance is good.

Key words: Fetomaternal hemorrhage syndrome, Machine learning algorithm, Prediction model, XGBoost, K-B test