Machine Learning-Based Predictive Modeling of Postpartum Depression.

Area of Science:

• Medical Informatics
• Computational Psychiatry
• Public Health

Background:

• Postpartum depression (PPD) is a significant maternal health concern.
• Current screening methods for PPD lack predictive capabilities for early intervention.

Purpose of the Study:

• To develop and evaluate machine learning (ML) models for predicting postpartum depression.
• To identify effective ML algorithms for PPD screening.

Main Methods:

• A retrospective cohort study utilized data from 28,755 women (2012-2013 PRAMS data).
• Balanced resampling techniques addressed data imbalance.
• Nine ML algorithms were trained and evaluated using 10-fold cross-validation.

Main Results:

• The Random Forest (RF) model achieved the highest accuracy (0.791) and AUC (0.884).
• Support Vector Machines (SVM) demonstrated strong performance with an AUC of 0.864.
• Model accuracies varied, with kNN showing the lowest at 0.650.

Conclusions:

• ML-based predictive modeling shows promise as a screening tool for postpartum depression.
• The developed models can aid in early identification and intervention for PPD.
• Further research can refine these ML approaches for clinical application.