# Artificial Intelligence-Based Prediction of Preeclampsia Using First-Trimester Biomarkers

## TL;DR

This study uses AI to predict preeclampsia in early pregnancy by combining patient data and biomarkers, showing high accuracy and potential for better prenatal care.

## Contribution

A novel AI framework using first-trimester biomarkers and machine learning to improve early preeclampsia detection.

## Key findings

- The deep neural network achieved 93.4% accuracy in predicting preeclampsia.
- PlGF, PAPP-A, and MAP were identified as key biomarkers for risk classification.
- The AI model outperformed traditional first-trimester algorithms like the FMF method.

## Abstract

Preeclampsia (PE) remains a leading cause of maternal and perinatal morbidity and mortality worldwide. Early identification of high-risk pregnancies during the first trimester is challenging, as traditional diagnostic methods based on maternal history and blood pressure readings often lack sensitivity. This study proposes an artificial intelligence (AI)-based predictive framework that integrates maternal demographic information, biophysical parameters, and first-trimester biochemical markers to enhance the early detection of PE. A curated dataset of first-trimester patient characteristics was used to develop and evaluate machine learning models, including support vector machines (SVM), random forests (RF), and deep neural networks (DNN). The AI framework demonstrated promising predictive performance, with the DNN achieving an accuracy of 93.4% on the held-out test set. Feature importance analysis identified placental growth factor (PlGF), pregnancy-associated plasma protein-A (PAPP-A), and mean arterial pressure (MAP) as key contributors to risk classification. While these results exceed the detection rates of traditional first-trimester algorithms such as the Fetal Medicine Foundation (FMF) algorithm, we have applied rigorous cross-validation, feature selection, and regularization techniques to mitigate overfitting. Future work will focus on external validation across multicenter cohorts and real-time clinical implementation to assess generalizability and clinical utility. Our findings suggest that AI-driven predictive analytics can support early risk assessment and personalized prenatal management, potentially improving maternal and fetal outcomes.