The study being conducted by Sibbald et al. aimed to determine, using machine learning models, how early postpartum depression (PPD) can be detected and if there are trimester-specific risk factors. Researchers used data from 2,865 Dutch pregnant women and they assessed biological, social, and psychological variables at 12, 20, and 28 weeks of gestation. These models were trained with increasing amounts of data where the first model included only first-trimester data, and the final model used variables collected throughout the entire pregnancy. PPD symptoms were studied 8-10 weeks postpartum using the Edinburgh Depression Scale (EDS). Research found that PPD prediction is possible as early as the first trimester, with accuracy and specificity being present across all machine learning models. However, sensitivity remained low, meaning that the models were better at identifying mothers who would not develop PPD rather than those who would. Additionally, psychological factors had the highest predictive accuracy, including depressive symptoms during pregnancy, negative affect, pregnancy distress, and previous mental health treatment. The most effective machine learning model was LassoLars used in the first and second trimesters. In sum, PPD can be predicted as early as 12 weeks using variables gathered from the first trimester, and this predictability is not drastically improved when including data from the second and third trimesters.