To evaluate the effectiveness of a machine learning model in predicting the short-term risk of preeclampsia using electronic health record data.
Key Findings:
Model performance improved through the third trimester, with strongest discrimination at approximately 34 weeks’ gestation.
High negative predictive values indicated strong ability to rule out near-term risk.
Positive predictive values improved as delivery approached, surpassing traditional risk-based models.
Blood pressure was the most influential predictor, with laboratory values more significant earlier in the third trimester.
Interpretation:
The machine learning model can dynamically estimate preeclampsia risk using routinely collected clinical data, potentially allowing for earlier interventions.
Limitations:
Retrospective design limits the findings.
Data was sourced from a single health system, which may affect generalizability.
Requires prospective validation before clinical implementation.
Conclusion:
The study suggests that machine learning models could enhance preeclampsia risk prediction and inform clinical decisions, but further validation is necessary.
In a target-trial emulation of more than 600,000 veterans, GLP-1 RA initiators saw fewer new substance use disorders—and patients with existing SUDs had fewer overdoses, hospitalizations, and deaths.
