Fig. 4: Loading plots of the proteins.

Sparse least squares model predicting high vs low air pollution exposure during pregnancy from maternal pregnancy week 24 proteomic profiles. A–C Overview of proteins with corresponding loadings. PM₁₀ only had one selected protein (AXIN1). The panels display linear regression estimates with 95% confidence intervals for proteins positively or negatively associated with air pollution measurements (n = 613). D–F The cross-validated predictions with AUC and p-values, which were strongly associated with air pollution. The box plots represent the class predictions for the median cross-validated model (n = 613). The center of the boxes represents the median, their bounds represent the 25th and 75th percentiles, and the lower and upper ends of whiskers represent the smallest and largest values.