Predicting Depression Onset in Pregnancy Using Machine Learning
Author Information
Author(s): Tamar Krishnamurti, Samantha Rodriguez, Bryan Wilder, Priya Gopalan, Hyagriv N. Simhan
Primary Institution: University of Pittsburgh
Hypothesis
Can machine learning algorithms predict the onset of first-time moderate-to-severe depression in pregnant individuals using early pregnancy data?
Conclusion
A small amount of self-reported data can accurately predict first-time depression among pregnant individuals.
Supporting Evidence
- The optimal model used only 14 variables and achieved an AUC of 0.89.
- Including food insecurity in the model improved its accuracy to an AUC of 0.93.
- The study found that 11.7% of participants experienced first-time depression during pregnancy.
Takeaway
Researchers used a smartphone app to collect information from pregnant people to help predict if they might get depressed later. This can help doctors give better care.
Methodology
Data was collected from 944 participants using a prenatal support app, where they reported clinical and social risk factors and completed depression screenings.
Potential Biases
The study may not adequately represent racially diverse groups and those with lower access to healthcare resources.
Limitations
Participation bias may limit generalizability, as the sample may not represent all pregnant individuals, particularly those with less access to technology.
Participant Demographics
78% identified as white, average age 30, 61% had income ≥ $50,000, 70% had a college degree or higher.
Statistical Information
P-Value
<0.01
Confidence Interval
[0.84, 0.94]
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website