A Prediction Model of Gestational Diabetes Mellitus Based on First Pregnancy Test Index.

The purpose of this study is to discuss the possibility of predicting gestational diabetes mellitus (GDM) by analyzing the first test indexes. In order to verify the prediction effect, we used 61 indexes, including age and 60 test indexes, from December 2015 to May 2018 in Beijing Pinggu District Hospital, and conducted experiments of GDM risk prediction based on a variety of different models, ranged from LR, LDA, RF to XGBoost. The experimental results reveal that compared to the dataset of using major relevant indicators, the dataset of using full indicators performs better. Besides, logistic regression can achieve a relatively good prediction effect. On the test set of all data, the area under the curve (AUC) of the Logistic regression model reaches 0.7787. In the meantime, the accuracy rate of the Logistic Regression model reaches (69.991 ± 2.833)%, and the recall rate and the mean value of the F1 value are (70.598 ± 2.210)% and (70.264 ± 2.128)%, respectively. So the analysis based on the first pregnancy test can play a role in predicting GDM to a certain extent.