AI Tool Predicts Postpartum Depression Risk at Birth with High Accuracy: Study

New Delhi: A new machine learning model developed by researchers at Mass General Brigham shows promising potential in predicting postpartum depression (PPD) risk using data available at the time of delivery. Published in the American Journal of Psychiatry, the study demonstrates how clinical and demographic information from electronic health records (EHR) can be used to identify individuals at higher risk for PPD, potentially allowing for earlier intervention and support.

Traditionally, PPD is screened for during postpartum visits, typically 6 to 8 weeks after childbirth, leaving many individuals without mental health support during the critical early weeks. The new model addresses this gap by analyzing factors such as medical history, demographics, and visit records—data already available in EHRs at the time of delivery.

To build and test the model, the team used data from 29,168 patients who gave birth between 2017 and 2022 across two academic medical centers and six community hospitals. Among them, 9 percent met the study’s criteria for PPD within six months post-delivery. The model effectively ruled out PPD in 90 percent of cases and identified nearly 30 percent of high-risk individuals who later developed the condition.

Importantly, the model performed consistently across age, race, and ethnicity, and even among those without prior psychiatric diagnoses. Incorporating prenatal scores from the Edinburgh Postnatal Depression Scale further improved accuracy.

"This is exciting progress toward developing a predictive tool that, paired with clinicians' expertise, could help improve maternal mental health," said lead author Mark Clapp, MD, MPH, of the Department of Obstetrics and Gynecology at Massachusetts General Hospital, a founding member of the Mass General Brigham healthcare system. "With further validation, and in collaboration with clinicians and patients, we hope to achieve earlier identification and ultimately improved mental health outcomes for postpartum patients."

Reference: https://www.massgeneralbrigham.org/en/about/newsroom/press-releases/machine-learning-model-helps-identify-patients-at-risk-of-postpartum-depression