Search

Background: Early identification of language disorders ensures children in need receive support to improve development and quality of life outcomes. Two previous studies identified six parent-reported survey items that predicted a comparable 11-year language outcome with ≥71% sensitivity and specificity. One used data from the Early Language in Victoria Study (ELVS) and the other from the Longitudinal Study of Australian Children (LSAC) birth cohort. This study will combine data from these two cohorts and add polygenic scores (PGS) of 24-38-month vocabulary. We will assess whether predictive accuracy of 11-year language outcome improves when adding PGS to a set of parent-reported behavioral predictors. Method: ELVS and LSAC collected six parent-reported measures about the child’s development between 24-36 months of age and a language outcome at 11 years. Both studies also collected genetic data at 7 (ELVS) or 11-12 (LSAC) years of age from which we derived PGS of 24-38-month vocabulary. We will use an ensemble machine-learning approach (SuperLearner) to estimate the accuracy of the six survey items and the PGS in predicting low 11-year language outcome (>1.5SD below the mean). Expected results: Predictive accuracy may improve when combining the parent-reported survey items with PGS. This will inform the extent to which future studies should prioritize collecting genetic data to help predict language outcomes.