You are viewing the site in preview mode

Skip to main content

Advertisement

Table 2 Area under the curve, accuracy, and the three most important predictors for the prediction of small for gestational age (SGA) birth using logistic regression and five machine learning methods pre-pregnancy and at 26 weeks in primiparous and multiparous women

From: Comparison of logistic regression with machine learning methods for the prediction of fetal growth abnormalities: a retrospective cohort study

  Pre-pregnancy 26 weeks
LR EN CT RF GB NN LR EN CT RF GB NN
SGA - Primiparae
 Area under the curve 0.592 0.598 0.569 0.601 0.609 0.600 0.662 0.661 0.627 0.650 0.665 0.660
 Accuracy 0.839 0.845 0.815 0.841 0.851 0.841 0.847 0.849 0.829 0.844 0.846 0.849
 Maternal age          
 Area-level income quintile             
 Pre-pregnancy smoking     
 Pre-pregnancy BMI        
 Pre-existing hypertension            
 Gravidity            
 Weight gain at 26 wks           
 Smoking in pregnancy           
 Pregnancy-induced hypertension            
SGA – Multiparae
 Area under the curve 0.741 0.744 0.711 0.715 0.728 0.741 0.771 0.771 0.713 0.745 0.766 0.772
 Accuracy 0.905 0.903 0.916 0.897 0.902 0.906 0.912 0.912 0.801 0.903 0.911 0.914
 Pre-pregnancy smoking         
 Pre-pregnancy BMI        
 Pre-existing hypertension             
 Previous LBW infant        
 Previous infant > 4080 g           
 Previous preterm delivery < 29 wks             
 Weight gain at 26 wks           
 Smoking in pregnancy             
 Pregnancy-induced hypertension            
  1. Abbreviations: BMI body mass index, CT classification tree, EN elastic net, GB gradient boosting, LBW low birth weight, LR logistic regression, NN neural network, RF random forest, wks weeks