image: This study included 12,715 pregnancies (with 140 cases of abnormal pregnancy excluded), among which 556 cases developed PE. Clinical data (age, BMI, plurality, parity, IVF, etc.), ultrasound parameters (UtA-PI), and laboratory markers (liver function, blood lipids, inflammation, trace elements, etc.) were collected before 14 weeks of gestation. Multivariate logistic regression, RCS analysis, and two-stage linear regression models were used to evaluate the association and threshold effects between laboratory markers and PE. Subsequently, the preterm PE and term PE cohorts were randomly divided into training and testing sets at an 8:2 ratio, respectively. The Boruta algorithm and seven machine learning models were employed for feature selection and the construction of laboratory markers prediction models. The optimal model was selected based on AUC and sensitivity, and SHAP values were used to quantify the contribution of features. Next, the prediction score of the optimal laboratory markers prediction model was integrated with the FMF model (maternal risk factors, MAP, PLGF, and UtA-PI) for the construction of the predictive modeling of PE, and the prediction performance was evaluated. Finally, cfDNA sequencing and TSS profiling were performed on 51 maternal plasma samples (18 cases of preterm PE and 33 controls), revealing early metabolic dysfunction and heightened inflammation in cases destined for PE. PE, preeclampsia; BMI, body mass index; IVF, in vitro fertilization; UtA-PI, uterine artery pulsatility index; RCS, restricted cubic spline; AUC, area under the receiver operating characteristic curve; SHAP, shapley additive explanations; FMF, fetal medicine foundation; MAP, mean arterial pressure; PLGF, placental growth factor; cfDNA, cell-free DNA; TSS, transcription start site; GGT, gamma-glutamyltransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; CHE, cholinesterase; FBG, fasting blood glucose; UAR MoM, uric acid-albumin ratio multiple of the median; TC, total cholesterol; TyG MoM, triglyceride-glucose index multiple of the median; CRP, c-reactive protein; Cu, copper; Ca, calcium; Mg, magnesium; Fe, iron; CatBoost, categorical boosting; LR, logistic regression.
Credit: Yi Zhu, Yanqiu Zhang, Chao Huang, Sheng Zhang, Jun Cao, Nicole Miranda, Sarina Zhao, Yan Peng , Chao Yu, Bin Feng, Jieyu Jin, Qingqin Tang, Jiaming Fan, Longwei Qiao, Yuting Liang
Preeclampsia (PE) is a serious pregnancy complication characterized by hypertension and multi-organ dysfunction after 20 weeks of gestation, posing significant risks to both mothers and infants. Early identification of women at high risk enables timely preventive interventions, yet established screening protocols, such as the Fetal Medicine Foundation (FMF) clinical model, exhibit moderate sensitivity during the first trimester. Since routine laboratory tests reflect early metabolic and inflammatory changes associated with disease development, they may provide an accessible and cost-effective strategy for improving PE prediction.
In a study published in Genes & Diseases, researchers from Soochow University, The University of Chicago Medical Center, Anhui Medical University, Nanjing Medical University, Chongqing Medical University, and Fudan University developed machine learning models integrating clinical risk profiles with laboratory biomarkers, using cell-free DNA (cfDNA) profiling to investigate underlying physiological alterations.
The researchers identified 26 laboratory indicators that differed significantly between women who later developed preeclampsia and those with uncomplicated pregnancies. These included markers of liver and kidney function, lipid metabolism, inflammation, electrolytes, and insulin resistance. Threshold analyses further demonstrated nonlinear relationships between several biomarkers—including gamma-glutamyltransferase (GGT), alkaline phosphatase (ALP), triglyceride-glucose (TyG) index, C-reactive protein (CRP), magnesium, iron, and high-density lipoprotein cholesterol (HDL-C)—and subsequent PE risk, suggesting that clinically meaningful cut-off values could improve individualized risk assessment.
Using Boruta feature selection and multiple machine learning algorithms, the investigators developed optimized prediction models for both preterm and term PE. For preterm disease, the AdaBoost model achieved the best performance, while a logistic regression model performed best for term PE. Importantly, combining routine laboratory markers with the existing FMF screening strategy significantly improved predictive accuracy. The integrated model achieved an AUC of 0.954 for preterm PE and 0.913 for term PE, substantially outperforming the FMF model alone and markedly increasing detection rates at the same false-positive threshold. SHAP analyses further identified inflammatory and metabolic markers, including UAR, TyG index, prealbumin, GGT, and CRP, as major contributors to prediction.
To investigate the biological basis of these findings, the researchers analyzed cell-free DNA promoter profiles and identified early transcriptomic alterations associated with immune dysregulation, oxidative stress, metabolic dysfunction, and abnormal placental angiogenesis. These observations support the concept that preeclampsia originates from systemic physiological disturbances well before clinical symptoms appear, rather than representing an isolated placental disorder.
Overall, this study demonstrates that integrating inexpensive first-trimester laboratory tests with machine learning substantially improves early prediction of both preterm and term preeclampsia. By combining routinely available clinical data with explainable artificial intelligence and mechanistic transcriptomic analyses, the findings provide a practical, cost-effective framework for earlier identification of high-risk pregnancies and may facilitate more personalized preventive strategies, particularly in resource-limited healthcare settings.
Reference
Title of Original Paper: First-trimester preeclampsia prediction model via integrative machine learning of maternal risk profiles and laboratory markers
Journal: Genes & Diseases
Genes & Diseases is a journal for molecular and translational medicine. The journal primarily focuses on publishing investigations on the molecular bases and experimental therapeutics of human diseases. Publication formats include full length research article, review article, short communication, correspondence, perspectives, commentary, views on news, and research watch.
DOI: https://doi.org/10.1016/j.gendis.2026.102075
# # # # # #
Genes & Diseases publishes rigorously peer-reviewed and high quality original articles and authoritative reviews that focus on the molecular bases of human diseases. Emphasis is placed on hypothesis-driven, mechanistic studies relevant to pathogenesis and/or experimental therapeutics of human diseases. The journal has worldwide authorship, and a broad scope in basic and translational biomedical research of molecular biology, molecular genetics, and cell biology, including but not limited to cell proliferation and apoptosis, signal transduction, stem cell biology, developmental biology, gene regulation and epigenetics, cancer biology, immunity and infection, neuroscience, disease-specific animal models, gene and cell-based therapies, and regenerative medicine.
Scopus Cite Score: 10.4 | Impact Factor: 14.6
# # # # # #
More information: https://www.keaipublishing.com/en/journals/genes-and-diseases/
Editorial Board: https://www.keaipublishing.com/en/journals/genes-and-diseases/editorial-board/
All issues and articles in press are available online in ScienceDirect (https://www.sciencedirect.com/journal/genes-and-diseases).
Submissions to Genes & Diseases may be made using Editorial Manager (https://www.editorialmanager.com/gendis/default.aspx).
Print ISSN: 2352-4820
eISSN: 2352-3042
CN: 50-1221/R
Contact Us: editor@genesndiseases.cn
X (formerly twitter): @GenesNDiseases (https://x.com/GenesNDiseases)
# # # # # #
Journal
Genes & Diseases