Mass General Brigham develops model to detect postpartum depression risk

Postpartum depression (PPD) affects up to 15% of individuals after childbirth, often leading to prolonged suffering before receiving help. Researchers at Mass General Brigham have created a machine learning model aimed at identifying patients at risk of PPD earlier in the postpartum period. This innovative model utilizes easily accessible clinical and demographic data from electronic health records (EHR) to assess PPD risk during delivery.

The study, published in the American Journal of Psychiatry, involved data from over 29,000 pregnant patients who delivered at two academic medical centers and six community hospitals within the Mass General Brigham system from 2017 to 2022. Among these patients, 9% developed PPD within six months post-delivery. By training the model on half of the patient data, researchers achieved a predictive accuracy that effectively ruled out PPD in 90% of cases.

The findings indicate that nearly 30% of individuals flagged as high risk by the model went on to develop PPD within the same six-month period. This performance is two to three times more effective than traditional population risk estimates. Additionally, the model demonstrated consistent predictive capabilities across various demographics, including race, ethnicity, and age.

To ensure it could accurately identify risks in low-risk patients, the study specifically excluded individuals with prior psychiatric diagnoses. The researchers also found that incorporating scores from the Edinburgh Postnatal Depression Scale collected during the prenatal period enhanced the model’s accuracy, suggesting that this established tool could be beneficial for both prenatal and postpartum assessments.

As the researchers prepare to test the model’s effectiveness in real-world settings, they are collaborating with healthcare providers and stakeholders to integrate the model into clinical practice. Dr. Roy Perlis, leading the project, emphasizes the importance of this work in developing a predictive tool that, alongside clinician expertise, could improve maternal mental health outcomes. The goal is to facilitate earlier identification of PPD, thus enabling timely intervention and support for new parents.