Boosting for postpartum depression prediction

Pregnancy and childbirth are important transitional life events for women. Like many other transitional life events, the effects of pregnancy and childbirth can have significant impact on a mother's physical and mental well-being. Sometimes they can even lead to Postpartum Depression (PPD). If left untreated, PPD can be debilitating for the mother and can adversely affect her ability to take care of herself and her infant. Since PPD is not clinically diagnosable, we consider the problem of predicting PPD from survey data about demographics, depression, and pregnancy etc. We adapt the successful functional-gradient boosting algorithm that can handle class imbalance in a principled manner. Our results demonstrate that the proposed machine learning approach can outperform the baseline classifiers and, consequently, demonstrate the potential of machine learning in predicting PPD.