PROJECT SUMMARY
Alcohol use and smoking during pregnancy often occur in the context of maternal anxiety and depression.
However, little is known regarding how different combinations of maternal alcohol use, smoking, and
psychiatric symptoms interact to alter early central and autonomic nervous system (CNS, ANS) development
and subsequent neurobehavioral outcomes in offspring. To better understand these relationships, this grant
proposes to comprehensively characterize maternal alcohol, smoking, depression, and anxiety patterns during
pregnancy via machine learning and artificial intelligence methodologies; to perform deep phenotyping of
electrocortical and physiological data from infants; to test the overall hypothesis that physiology-derived
measures mediate the relationship between maternal prenatal exposures and subsequent child
neurodevelopmental outcomes. To accomplish these goals, this research proposes to use existing data from
the completed Safe Passage Study conducted by the NIH funded Prenatal Alcohol in SIDS and Stillbirth
(PASS) Network and from recently completed follow-up studies which enrolled subsets of the PASS
participants. PASS cohort participants are from geographically and socio-economically diverse regions in
South Africa and the Northern Plains of the United States. The extant PASS database comprises a detailed
documentation of daily alcohol and smoking exposure and prenatal depression/anxiety assessments collected
for 10,325 maternal/infant dyads and more than 1,000 variables which provide extensive information on
maternal-fetal-infant medical history, demographics, and socio-economic status. These exposure data will be
analyzed in relation to neonatal CNS/ANS deep phenotyping profiles derived from the analysis of 3,355 infant
electroencephalograms (EEG) and electrocardiogram (ECG) data recorded either at birth or at one month of
age. The extant neurobehavioral outcomes data were collected in four distinct cohorts of participants originally
enrolled in PASS and followed up at 12 months (n=849) and 24-37 months of age (n=667). The follow-up study
datasets contain item-level data, sub-test raw scores, and scaled-scores for the Mullen scales of early learning,
the Parent Report of Children’s Abilities-Revised (PARCA-R), the Bayley scales of infant development III
screening test and the Modified Checklist for Autism in Toddlers Revised with Follow up (M-CHAT-R/F). The
proposed approach is expected to elucidate mechanisms of suboptimal neurodevelopment and resilience, and
ultimately inform screenings for at-risk groups. Additionally, the proposed comprehensive deep phenotyping is
intended to identify markers that are potential mediators of the relationship between prenatal exposures and
outcomes. We anticipate that our findings will capitalize on the geocultural diversity of our cohorts and provide
meaningful information associated with health disparities in underrepresented populations.
