Prediction Model of Adverse Pregnancy Outcome in Pre-Eclampsia Based on Logistic Regression and Random Forest Algorithm.

AIM: To construct a prediction model for adverse pregnancy outcomes of preeclampsia (PE). Thus assisting clinicians to identify high-risk patients. Provide guidance for treatment intervention.

METHODS

A retrospective study was conducted on 319 PE patients admitted to the Huzhou Maternal and Child Health Hospital from April 2021 to December 2022, The patients were divided into an adverse group (93 cases) and a non-adverse group (226 cases) based on whether they had adverse pregnancy outcomes after admission. Collect clinical data from patients, using a single factor analysis to screen statistically significant indicators as input variables, the outcome of the analysis is dependent on the incidence of PE adverse pregnancy outcomes. Divide patients into training and testing sets in a 7:3 ratio, Logistic regression model and random forest model were constructed respectively. Evaluate the predictive performance of two statistical models.

RESULTS

Among the 319 PE patients included 93 had adverse pregnancy outcomes after admission. Among them, Age (OR: 1.702, 95%CI: 1.069~2.710), small gestational age (OR: 0.757,95%CI: 0.607~0.945), more clinical symptoms (OR: 3.618, 95%CI: 1.682~7.783), high 24 h proteinuria (OR: 2.532, 95%CI: 1.290~4.968), low PLT index (OR: 0.616, 95%CI: 0.419~0.906), high AST index (OR: 1.554, 95%CI: 1.012~2.387), high D-Dimer index (OR:1.966, 95%CI: 1.183~3.267) were the influencing factors of adverse pregnancy outcomes in PE patients. The test set found that the random forest model was superior to the Logistic regression model in predicting the risk of adverse pregnancy outcomes in PE patients.

CONCLUSIONS

The random forest model has good stability in predicting the risk of adverse pregnancy outcomes in PE, and its prediction efficiency is better than the Logistic regression model.