Preeclampsia risk model: 18-month retrospective validation

Materna's preeclampsia risk algorithm, running every hour against the latest BP, weight, urine protein, and history features, achieved 0.84 AUROC on a retrospective cohort of 4,200 pregnancies.

False-positive rate was held below 8% to keep alert fatigue low. The model is intentionally framed as decision support, not a substitute for clinical judgment.

The full methodology, feature set, and threshold tuning protocol are available on our research page.