Abstract:
Background: Early life factors, including parental sociodemographic characteristics,pregnancy exposures, and physical and neurodevelopmental features measured in infancy are associated with fetal alcohol spectrum disorders (FASD). The objective ofthis study was to evaluate the performance of a classifier model for diagnosing FASDin preschool-aged children from pregnancy and infancy-related characteristics.Methods: We analyzed a prospective pregnancy cohort in Western Ukraine enrolledbetween 2008 and 2014. Maternal and paternal sociodemographic factors, mater-nal prenatal alcohol use and smoking behaviors, reproductive characteristics, birthoutcomes, infant alcohol-related dysmorphic and physical features, and infant neu-rodevelopmental outcomes were used to predict FASD. Data were split into separatetraining (80%: n = 245) and test (20%: n = 58; 11 FASD, 47 no FASD) datasets. Trainingdata were balanced using data augmentation through a synthetic minority oversampling technique. Four classifier models (random forest, extreme gradient boosting[XGBoost], logistic regression [full model] and backward stepwise logistic regression)were evaluated for accuracy, sensitivity, and specificity in the hold-out sample.Results: Of 306 children evaluated for FASD, 61 had a diagnosis. Random forest mod-els had the highest sensitivity (0.54), with accuracy of 0.86 (95% CI: 0.74, 0.94) inhold-out data. Boosted gradient models performed similarly, however, sensitivity wasless than 50%. The full logistic regression model performed poorly (sensitivity = 0.18and accuracy = 0.65), while stepwise logistic regression performed similarly to theboosted gradient model but with lower specificity. In a hold-out sample, the bestperforming algorithm correctly classified six of 11 children with FASD, and 44 of 47children without FASD.Conclusions: As early identification and treatment optimize outcomes of childrenwith FASD, classifier models from early life characteristics show promise in predicting FASD. Models may be improved through the inclusion of physiologic markers ofprenatal alcohol exposure and should be tested in different samples.
