Zhang, Y., Wang, S., Hermann, A., Joly, R., & Pathak, J. (2021). Development and validation of a machine learning algorithm for predicting the risk of postpartum depression among pregnant women. Journal of affective disorders, 279, 1-8.
Intervention Components (click on component to see a list of all articles that use that intervention): , EMR Reminder, Targeting Interventions to Focused Groups, Educational Material, Mobile Apps, Online Material/Education/Blogging
Intervention Description: The study primarily focuses on developing a data-driven primary intervention approach using machine learning and electronic health records (EHR) data to identify pregnant women at risk for postpartum depression (PPD) . The intervention aligns with a discernible strategy of leveraging machine learning algorithms to predict PPD risk based on EHR data, with the potential for early prevention, diagnosis, and intervention . The study does not analyze a multicomponent intervention; rather, it focuses on the development and validation of a machine learning algorithm for PPD risk prediction using EHR data. The intervention strategy is centered around leveraging data-driven approaches to identify at-risk individuals and potentially tailor therapeutic interventions, screening timelines, and preventive strategies for PPD
Intervention Results: The study analyzed a total of 15,197 deliveries from January 2015 to June 2018, and the prevalence of depression was 6.7% (N=1,010) and 6.5% (N=3,513) in the WCM and NYC-CDRN datasets, respectively . The machine learning algorithm was able to predict PPD risk with an area under the receiver operating characteristic curve (AUC-ROC) of 0.83 (95% CI: 0.81-0.85) in the training dataset and 0.80 (95% CI: 0.77-0.83) in the validation dataset . The study found significant differences in age, the number of emergency department visits, and racial distribution between PPD and non-PPD groups in the training and validation data . The study demonstrates that a data-driven primary intervention approach using machine learning and EHR data may be leveraged to reduce the healthcare provider burden of identifying PPD risk.
Conclusion: Machine learning-based models incorporating EHR-derived predictors, could augment symptom-based screening practice by identifying the high-risk population at greatest need for preventive intervention, before development of PPD.
Study Design: The study design was a prospective cohort study that used electronic health records (EHR) data to develop and validate a machine learning algorithm for predicting the risk of postpartum depression (PPD) among pregnant women . The study used two EHR datasets containing data on 15,197 women from 2015 to 2018 at a single site and 53,972 women from 2004 to 2017 at multiple sites as development and validation sets, respectively . The study included all pregnant women with fully completed antenatal care procedures who had live births of infants, and the exclusion criteria were maternal age below 18 or above 45, or lack of outpatient, inpatient, or emergency room encounter information in the EHR data within 1 year following childbirth . The study was approved by the Institutional Review Board at Weill Cornell Medicine (IRB protocol# 1711018789), and data extraction and analysis were performed in 2019 . The study used a well-defined outcome measure of PPD diagnosis within 1 year following childbirth, and the machine learning algorithm was able to predict PPD risk with a high degree of accuracy
Setting: The study setting for the development dataset was a single site, and the validation dataset included data from multiple health systems across New York City affiliated with the Patient-Centered Outcomes Research Institute funded New York City Clinical Data Research Network data (NYC-CDRN) . Therefore, the study setting primarily involved healthcare institutions and systems in New York City, USA.
Population of Focus: The target audience for the study includes healthcare professionals, researchers, and policymakers involved in maternal and mental health, as well as professionals working with electronic health records (EHR) and machine learning applications in healthcare. Additionally, the findings of the study may be of interest to organizations and institutions involved in developing and implementing predictive models for identifying and addressing the risk of postpartum depression among pregnant women.
Sample Size: The study included a total of 15,197 deliveries from January 2015 to June 2018 in the development dataset, and 53,972 deliveries from August 2004 to October 2017 in the validation dataset . These datasets were used to develop and validate a machine learning algorithm for predicting the risk of postpartum depression among pregnant women.
Age Range: The study included pregnant women within a specific age range. The exclusion criteria for the study were maternal age below 18 or above 45 . Therefore, the age range of the included pregnant women in the study was 18 to 45 years old.
Access Abstract