Artificial intelligence predicts pregnancy complications based on cytokine profiles

Area of Science:

• Reproductive immunology
• Computational biology
• Biomedical data science

Background:

• Early prediction of pregnancy complications is vital for effective prevention and management.
• Maternal and fetal health outcomes are significantly impacted by timely interventions.
• Understanding predictive biomarkers can reduce pregnancy-related pathogenesis.

Purpose of the Study:

• To investigate the prognostic value of cytokines as predictors of pregnancy complications.
• To apply unbiased artificial intelligence/machine learning (AI/ML) methods for predictive modeling.
• To differentiate between normal delivery and various pregnancy complications using cytokine profiles.

Main Methods:

• Analysis of seven cytokines from activated peripheral blood mononuclear cells (PBMC) in 127 women with complications and 97 controls.
• Application of AI/ML algorithms including kNN, SVM, decision tree, and ensemble classification.
• Binary classification tasks to predict conditions like recurrent spontaneous miscarriage (RSM), preterm delivery (PTD), pregnancy-induced hypertension (PIH), and premature rupture of fetal membranes (PROM).

Main Results:

• Significant differences in IL-2 and IFN-γ cytokine levels were observed across various pregnancy conditions.
• High accuracies and f-measures were achieved using Ensemble (Bagged), QDA, and SVM (Cubic) classifiers.
• AI/ML models effectively distinguished normal delivery from pregnancy complications.

Conclusions:

• A novel AI/ML-based methodology for predicting pregnancy complications using cytokine levels has been developed.
• Cytokine profiles from peripheral blood cells serve as valuable biomarkers for early detection.
• This approach supports timely clinical interventions to improve maternal and fetal outcomes.