A Comprehensive First-Trimester Predictive Model for Preeclampsia Based on Multi-Indicators and Machine Learning: A Retrospective Single-Center Study

Haixia LiangXuejing ZhaoYing ZhangYujie WuHan WuZehui ZhangHe Ying^*

• Xijing hospital the 986th hospital department,The Fourth Military Medical University, Shanxi xi’an, China

Background: Preeclampsia (PE) is a severe, unpredictable disorder contributing to maternal and perinatal morbidity and mortality. This study develops a comprehensive first-trimester predictive model for PE using maternal, biophysical, biochemical, and hematological indicators Methods: This retrospective study included 100 pregnant individuals with singleton gestations (50 PE, 50 controls). Various early pregnancy indicators, including hematological, biochemical, inflammatory, angiogenic, and biophysical markers, were collected. LASSO regression was used for feature selection. Subsequently, seven different machine learning algorithms were employed for model development. Model performance was evaluated using Receiver Operating Characteristic (ROC) curves. An independent external validation cohort of 70 participants (35 PE, 35 controls) was used to confirm the model's generalizability. Results: Baseline characteristics showed significantly higher early pregnancy systolic blood pressure and diastolic blood pressure (SBP/DBP) in the PE group (P<0.001). Early pregnancy indicator comparisons revealed the PE group had significantly higher median WBC count, neutrophil count, monocyte count, and C-reactive protein (CRP) levels, and lower median hemoglobin (Hb) and hematocrit (HCT). Derived indices like the neutrophil-to-lymphocyte ratio (NLR) were significantly higher (P<0.001). Crucially, Placental Growth Factor (PlGF) levels were significantly lower (P<0.001), while uterine artery pulsatility index (PI) was significantly higher (P<0.001). LASSO regression identified 12 key predictive features, including PlGF, uterine artery PI, CRP, and NLR. Among the machine learning models, the Neural Network (NNET) model demonstrated the highest predictive performance, with an Area Under the Curve (AUC) of 0.917. The model maintained strong performance (AUC = 0.838) in external validation. SHAP analysis confirmed PlGF, uterine artery PI, CRP, and NLR as the most influential features. Conclusion: We developed a robust predictive model for PE based on early pregnancy biomarkers and machine learning techniques. The NNET model demonstrated superior discriminative ability in both internal and external validation cohorts. Early identification of high-risk pregnancies using this model could facilitate timely interventions, such as low-dose aspirin, potentially improving maternal and fetal outcomes. Further multi-center prospective studies are warranted to validate the model on a broader scale.