Background: Preeclampsia (PE) is a complex disorder with significant maternal and fetal risks. The soluble fms-like tyrosine kinase-1 (sFlt-1) and placental growth factor (PlGF) ratio shows promise as a diagnostic tool, but its adoption in the U.S. is limited, due to a lack of widely available testing platforms, locally conducted studies based on the U.S. population and clear, evidence-based guidance on test implementation, including appropriate cutoff values. Patients/Materials and Methods: This is a cohort study evaluating the sFlt-1/PlGF ratio for predicting PE within two weeks among pregnant individuals ([≥]18 years old, gestational age [≥]20 weeks. Serum samples were collected during routine prenatal visits or using leftover samples from triage evaluations. Patients diagnosed with PE before sample collection, those with postpartum samples, or those missing delivery or diagnostic data were excluded. sFlt-1/PlGF ratios were measured using Roche Elecsys assays, and predictive performance was assessed by operating characteristics (ROC) curve analysis with logistic regression, bootstrapping, and cross-validation. Results: The sFlt-1/PlGF ratio was significantly higher in hypertensive patients (n = 119) than in non-hypertensive patients (n = 346) (p < 0.001). PE developed in 9.7% of all patients and 37.0% of hypertensive patients within two weeks. ROC analysis showed an area under the curve (AUC) of 0.92 (95%CI: 0.88-0.95) for the overall population, higher than 0.82 (95%CI: 0.73-0.89) in the hypertensive group. The optimal cutoff for the overall population was 33 (95% CI: 26-47), yielding a sensitivity of 93.8% (95%CI: 85.4-100.0), specificity of 81.3% (95%CI: 76.8-85.8), negative predictive value (NPV) of 99.2% (95%: 98.0-100.0), and positive predictive value (PPV) of 35.3% (95%CI: 25.1-45.5). For the hypertensive group, the optimal cutoff was 50 (95% CI: 26-84), with a sensitivity of 81.7% (95%: 61.8-97.4), specificity of 73.1% (95%CI: 53.3-90.4), NPV of 87.6% (95%CI: 76.6-97.7), and PPV of 65.2% (95%CI: 50.0-82.5). Conclusions: The sFlt-1/PlGF ratio shows high NPVs for ruling-out PE, but moderate PPVs, limiting its rule-in ability. Additionally, our findings indicate different performance in the overall and hypertensive patients, highlighting the need for further research to refine diagnostic thresholds and improve clinical guidance.
The Hidden Power of Normalization: Exponential Capacity Control in Deep Neural Networks
arXiv:2511.00958v1 Announce Type: cross Abstract: Normalization methods are fundamental components of modern deep neural networks (DNNs). Empirically, they are known to stabilize optimization dynamics and