Related Experiment Video
Updated: Jun 26, 2025
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Development and Validation of an Automated, Real-Time Predictive Model for Postpartum Hemorrhage.
Holly B Ende1, Henry J Domenico, Aleksandra Polic
1Departments of Anesthesiology, Biostatistics, Obstetrics and Gynecology, and Biomedical Informatics, Vanderbilt University Medical Center, and Vanderbilt University School of Medicine, Nashville, Tennessee; and the Department of Anesthesiology, Boston Children's Hospital, Boston, Massachusetts.
A new postpartum hemorrhage prediction model using electronic health record data shows superior performance compared to existing tools. This validated model can be integrated into clinical care for real-time risk assessment.
Area of Science:
- Obstetrics and Gynecology
- Clinical Informatics
- Health Services Research
Background:
- Postpartum hemorrhage (PPH) is a leading cause of maternal morbidity and mortality.
- Accurate and timely risk prediction is crucial for effective PPH management.
- Existing risk assessment tools may lack sufficient accuracy or real-time applicability.
Purpose of the Study:
- To develop and validate a predictive model for PPH using automated, real-time electronic health record (EHR) data.
- To compare the performance of the developed model against a nationally recognized risk prediction tool.
- To facilitate clinical deployment through EHR integration.
Main Methods:
- A multivariable logistic regression model was developed using retrospective EHR data from 21,108 deliveries.
- The model was derived and validated using an 80/20 split, with PPH defined as ≥1,000 mL blood loss plus transfusion.
- Performance was assessed using the area under the receiver operating characteristic curve (AUC) and compared to the CMQCC tool; prospective validation was also performed.
Main Results:
- The predictive model demonstrated strong discrimination with an AUC of 0.81 in the derivation set and maintained performance in temporal validation (AUC 0.80) and prospective validation (AUC 0.82).
- The developed model significantly outperformed the CMQCC tool (AUC 0.69, P <.001).
- The model showed excellent calibration and was successfully implemented and validated in real-time using EHR data.
Conclusions:
- A robust, temporally validated PPH prediction model was developed and integrated into the EHR.
- The model exhibits superior predictive performance compared to a widely used risk assessment tool.
- Real-time validation confirms its utility for clinical decision support in PPH prevention and management.
