Background <p>Few studies have explored the associations between early pregnancy laboratory tests and preeclampsia, leaving rich first-trimester data underutilized. This study introduces a pipeline, integrating Generalized Additive Models with dose-response analysis, to elucidate the complex associations.</p> Methods <p>Our pipeline includes identifying turning points (pathway A) and risk/protective intervals (pathway B). We used Generalized Additive Models with cubic regression splines. For pathway A, turning points were identified at the second derivative extrema of nonlinear monotonic fitted probability curves. For pathway B, risk/protective intervals were determined where the Odds Ratios equaled 1. Propensity score matching and the risk ratios were used for evaluation. This pipeline was applied to a retrospective GZ cohort of 12,474 pregnancies and evaluated in an external GG cohort.</p> Results <p>Through analyzing 99 unique laboratory tests within GZ cohort, our pipeline highlighted 16 exhibiting optimal turning points via pathway A and 4 showing risk/protective intervals through pathway B. The turning points from pathway A were comparable to those from traditional piecewise logistic regression. Evaluation within GG cohort confirmed the statistical robustness. Additionally, our experiments demonstrated that hyperparameter fine-tuning of Generalized Additive Models fitting had minimal effect, and pathway output metrics are sensitive to pregnancy stages, leading to considerable variability in conclusions.</p> Conclusions <p>The proposed pipeline was rigorously validated for its efficacy across two independent cohorts, achieving consistent outcomes. Furthermore, the laboratory tests identified mostly align with conclusions from prior studies. We believe it both advances our understanding of the mechanisms during early pregnancy disorder and offers vital insights for early preeclampsia detection and preventive interventions.</p> Trail registration <p>NA.</p>