Gestational Dating by Urine Metabolic Profile at High Resolution Weekly Sampling Timepoints

The study characterizes the weekly baseline profile of the human pregnancy metabolome of urine samples from two normal pregnancy cohorts (discovery: 478 samples from 19 subjects at California; validation: 171 samples from 10 subjects at Alabama). Results: 5,473 urinary metabolic features were identified by LC-MS untargeted metabolomics. Prediction models (on six urine metabolites) were constructed based on gradient-boosting decision trees to date gestational age in high accordance with ultrasound results, and to accurately predict time to delivery.