Comparison of Two Mathematical Models for the Screening of Preeclampsia Using Free Cloud Computing Resources

Abstract

Aim: To develop and validate a free computational tool for preeclampsia (PE) risk assessment and compare the performance of two widely used mathematical screening models: the Fetal Medicine Foundation (FMF) competing risks model and the Fetal Medicine Barcelona (FMB) multivariate Gaussian distribution model. Methods: A Python-based computational engine was developed using Google Colab, which integrated maternal characteristics, mean arterial pressure (MAP), uterine artery pulsatility index (UtA-PI), and placental growth factor (PlGF). Simulated cohorts of 100,000 pregnancies (97,000 non-PE and 3,000 with PE) were assessed. Concentrations of PlGF were simulated across three analytical platforms (Roche, Thermo Fisher, and Perkin Elmer) at 11–13 weeks of gestation (WG) and with a Roche platform at 10 WG. The model performance was evaluated using Receiver Operating Characteristic (ROC) curve analysis, detection rates (DRs), and screen-positive rates (SPRs). Results: The computational tool showed excellent agreement with validated online calculators (proportional differences of 2.1% for FMF and 5.7% for FMB). The FMF model consistently outperformed the FMB model across all platforms (AUC, 0.885–0.888 vs. 0.845–0.846). Platform-specific PlGF differences significantly affected the risk thresholds but not the overall diagnostic accuracy. Both models maintained comparable performance at 10 WG. Conclusion: The FMF model outperformed the FMB model owing to the broader integration of maternal risk factors and platform-specific medians. This free, open-access tool supports informed PE screening decisions and is particularly relevant given the widespread commercial use of both models.