AUTHOR=Villalaín Cecilia , Herraiz Ignacio , Domínguez-Del Olmo Paula , Angulo Pablo , Ayala José Luis , Galindo Alberto 

TITLE=Prediction of Delivery Within 7 Days After Diagnosis of Early Onset Preeclampsia Using Machine-Learning Models

JOURNAL=Frontiers in Cardiovascular Medicine

VOLUME=Volume 9 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2022.910701

DOI=10.3389/fcvm.2022.910701

ISSN=2297-055X

ABSTRACT=Early-onset preeclampsia (eoPE) is a hypertensive disorder of pregnancy with endothelial dysfunction manifested before 34 weeks. In absence of end-organ damage or fetal distress, expectant management is recommended. However, disease progression is unpredictable and timing of hospitalization, corticosteroids and delivery remains a challenge. We aim to develop a prediction model using machine-learning tools for the need of delivery within 7 days of diagnosis (model D) and the risk of developing hemolysis, elevated liver enzymes, and low platelets (HELLP) syndrome or abruptio placentae (model HA).
Retrospective cohort of singleton pregnancies with eoPE and attempted expectant management between 2014 and 2020. A Mono-objective Genetic Algorithm based on supervised classification models was implemented to develop D and HA models. Maternal basal characteristics and data gathered during PE diagnosis: gestational age, blood pressure, platelets, creatinine, transaminases, angiogenesis biomarkers (soluble fms-like tyrosine kinase-1, placental growth factor) and fetal ultrasound data were pooled for analysis. The most relevant variables were selected by using bio-inspired algorithms. In order to cover different resource availability scenarios, we developed basal models that solely included demographic characteristics of the patient(D1, HA1), as well as advanced models adding information available at diagnosis of eoPE(D2,HA2). 
A total of 215 women with eoPE were evaluated and 47.9% required delivery within 7 days of diagnosis. Median time-to-delivery was 8 days. Basal models were better predicted by KNN in D1 which had a diagnostic precision of 0.68+/-0.09, with sensitivity (Sn) of 63.6% and specificity (Sp) of 71.4% using 13 variables and HA1 of 0.77+/-0.09, Sn of 60.4% and Sp of 80.8%. Models at diagnosis were better developed by SVM using 18 variables, where D2’s precision improved to 0.79+/-0.05 with 77.3%Sn and 80.1%Sp and HA2 had a precision of 0.79+/-0.08 with 66.7%Sn and 82.8%Sp.
At the time of diagnosis of eoPE, support-vector-machine classification model with evolutionary feature selection process provides good predictive information of the need for delivery within 7 days and development of HELLP/abruptio placentae, using maternal characteristics and markers that can be obtained routinely. This information could be of value when assessing hospitalization and timing of antenatal corticosteroids administration.