Fig. 4

Performance and gene analysis of the model in predicting healthy population and preeclampsia patients. A LASSO for gene selection. The vertical dotted line shows the best lambda value of 0.0029 selected through fivefold cross-validation. B Differentially expressed genes between preeclampsia placenta and normal placenta. *P-value < 0.001, t-test. C Overlapping genes between molecular markers of placenta subpopulations and preeclampsia pathology. D ROC curve of preeclampsia risk assessment model. E KS curve for preeclampsia risk score card. F, G Based on the ensemble model, the clustering effect of the LASSO optimal gene set and all genes is compared (F is the optimal LASSO optimal genes, G is all genes). H The importance of genes identified by different preeclampsia risk models, and the size of the circle represents the value of relative importance