出産後3ヶ月の育児ストレス予測における一般化加法モデルの使用

The aim of this study was to predict and explain parenting stress at 3 months after the delivery with pregnancy-related features collected in different stages of pregnancy using generalized additive modelling. Pregnant women (n=1138) were recruited for this study. The participants completed multiple questionnaires at gestational weeks 14, 24, 34, and postnatally. The Swedish Parenthood Stress Questionnaire, completed at 3 months postpartum, served as the primary outcome measure. All pregnancy-measured variables were employed as explanatory factors, and a machine learning model underwent training and testing. Initially, the model incorporated variables from gestational week 14, wherein factors associated with the mother's personal well-being-such as quality of life and depressive symptoms-predicted parenting stress. Subsequently, the model expanded to include variables from gestational weeks 14 and 24, and based on that, also factors related to the relationship, such as avoidance and anxiety in close relationships and satisfaction in the relationship, predicted postpartum parenting stress. As the model evolved further, encompassing variables from gestational weeks 14, 24, and 34, it revealed that factors pertaining to both relationship satisfaction and the mother's well-being continued to predict parenting stress. The study underscores the significance of addressing both maternal well-being and relationship issues during maternity care.