Zhifei Zeng ’23
Current research suggests that factors such as socioeconomic deprivation, inadequate prenatal care, unplanned pregnancy, and psychosocial vulnerability such as stress may contribute to prenatal depression. PROMOTE is a newly developed screening tool that identifies psychosocial vulnerability in prenatal populations by assessing social determinants of health, social resources, stress and health behaviors. A research group led by Heidi Preis of Stony Brook University used data from PROMOTE to quantify individual patients’ risk of prenatal depression and to develop machine learning (ML) methods to determine which factors in patients’ lives make them more vulnerable.
The researchers administered questionnaires to all female patients attending an outpatient antenatal clinic in a university hospital. PROMOTE was used to assess 19 background risk factors such as unintended pregnancy, employment, educational attainment, financial status, and substance use. The Edinburgh Postnatal Depression Scale (EPDS) was used to assess prenatal or postnatal depression. Participants were asked to rate the frequency of ten statements about feelings and thoughts experienced in the past two weeks. Scores were calculated by summing the scores (0 to 30) for each statement, with higher total scores indicating a greater risk of depression. The ML algorithm developed by the researchers classified patients as being at high or low risk for depression based on PROMOTE and EPDS data. The researchers also assessed the effect of each PROMOTE risk factor on the probability of depression risk. It was found that the three most important factors in PROMOTE for predicting depression risk were perceived stress, emotional problems, and lack of family support. Moreover, predictions of the risk of prenatal depression using the PROMOTE data ML algorithm agreed with the results of the EPDS. This suggests that the ML algorithm can help doctors predict the likelihood of prenatal depression.
The ML method developed by Stony Brook research group can be applied to clinical care to identify at-risk patients and detect which contextual factors contribute to their risk, enabling improved interventions and care delivery. Overall, this new application of machine learning allows healthcare practitioners to anticipate a patient’s risk of prenatal depression and provide personalized psychological interventions to patients based on their specific risk factors.
 H. Preis, et al., Applying machine learning methods to psychosocial screening data to improve identification of prenatal depression: implications for clinical practice and research. Archives of Women’s Mental Health 25, (2022). doi: 10.1007/s00737-022-01259-z.
 Image retrieved from: