Am J Psychiatry

APA 2025: Machine learning model predicts postpartum depression risk before hospital discharge

May 27, 2025

A machine-learning model enabled risk stratification for postpartum depression (PPD) at the time of hospital discharge, potentially allowing for individualized postpartum care planning and targeted interventions in resource-limited settings. The findings were shared at the American Psychiatric Association annual meeting.

Study details: A retrospective cohort study analyzed 29,168 individuals delivering at two academic and six community hospitals (2017–2022), excluding those with prior depression. An elastic net machine learning model was developed and externally validated to predict PPD within six months postpartum, using sociodemographic, medical, and prenatal depression screening data available before discharge. PPD was defined by mood disorder diagnosis, antidepressant prescription, or a positive Edinburgh Postnatal Depression Scale (EPDS) screen.

Results: Within six months postpartum, 9.2% of patients met criteria for PPD. The model demonstrated good discrimination, with an area under the receiver operating characteristic curve of 0.721 (95% confidence interval [CI] 0.709–0.736) and calibration (Brier score 0.087; 95% CI 0.083-0.091). At 90% specificity, the positive predictive value was 28.8% and the negative predictive value was 92.2%.

Source:

Clapp MA, et al. (2025, May 19). Am J Psychiatry. Stratifying Risk for Postpartum Depression at Time of Hospital Discharge. https://pubmed.ncbi.nlm.nih.gov/40384019/