ABSTRACT The current state of genomic diagnostics allows us to diagnose genetic disease early in life either by 1) identifying carrier status for inherited diseases before conception or 2) diagnosing a genetic condition during pregnancy. Genetic diagnosis at these two timepoints is inconsistently offered and without early diagnosis, early treatment including new in utero therapies may not achieve their full potential. The central hypothesis of this grant is that newly developed strategies can improve risk stratification and identify appropriate candidate individuals for early genetic testing in order to achieve earlier diagnosis and allow for earlier treatment. Aim 1 will reevaluate which inherited genetic conditions warrant universal preconception carrier screening. Preconception carrier screening guidelines for inherited diseases can be refined in such a way as to resolve conflicting guidelines from various medical societies, add important actionable genes that have early treatment available, and ultimately improve uptake of this available testing. In order to achieve this, two biobanks (that improve upon prior work in their increased sample size and diversity) will be used to update carrier frequencies for the genes previously recommended for universal carrier screening. Expert mentorship and didactic work in genomic epidemiology and large biobank dataset utilization will accomplish this aim and work toward building a genomic medicine public health research program. Aim 2 will use pregnancy health factors to identify fetuses and neonates at higher risk for genetic disease. Pregnancy health factors such as preeclampsia, gestational diabetes, and placental abnormalities, may serve as indicators that a pregnancy is high risk to be affected by a fetal genetic disease. Utilizing these factors in a machine learning model will enable risk stratification of a pregnancy into low, moderate, or high-risk for fetal genetic disease. The machine learning model will be built using a prior studied cohort of 236 pregnancies affected by fetal genetic diseases and 472 unaffected pregnancies. The model will be validated using a cohort of 325 mother-infant dyads with suspicion for genetic disease; 20 dyads in the validation cohort will be consented for return of results. Expert mentorship and didactics in statistical modeling and machine learning will accomplish this aim and work toward a research program that utilizes machine learning with existing and available health data to advance the diagnosis of rare disease on a public health scale. The aims of this proposal coupled with a mentorship and training plan will establish an independent research program focused on genomic epidemiology and machine learning to launch public health measures that advance the diagnosis and treatment of genetic disease in early life.