jenniferfedor / ms-thesis

MS thesis project using GAMM and elastic net regression to predict quality of life in young adulthood from individual differences in adolescent neurocognitive development

ms-thesis

Data preprocessing, cleaning, and analysis code for Biostatistics MS thesis work using GAMM and elastic net regression to predict quality of life in young adulthood from individual differences in adolescent neurocognitive development

Abstract

Background: Inhibitory control and the brain systems that support this cognitive process are known to undergo a protracted maturation through adolescence. Relatively little work, however, has examined how individual differences in the normative adolescent development of inhibitory control may be associated with outcomes such as health-related quality of life (QOL) in early adulthood.

Methods: We analyzed data from an accelerated longitudinal study of healthy individuals initially aged 8-30 who, at approximately yearly intervals, completed an inhibitory control task while functional magnetic resonance imaging data were acquired. Generalized additive mixed models were utilized to characterize age-related change in inhibitory control behaviorally and in regional brain activation. Random intercepts and slopes from these models, representing person-specific deviations from group-level developmental trajectories, were utilized as covariates in a bootstrap- enhanced elastic net regression procedure to predict QOL, which was assessed with a self-report questionnaire at the study endpoint.

Results: There were significant developmental improvements in inhibitory control behaviorally that continued into young adulthood. Among examined motor response and executive control brain regions, there were significant age-related decreases in activation during correctly performed task trials occurring until mid-adolescence in the L frontal eye fields (FEF), bilateral posterior parietal cortex (pPC), and R dorsolateral prefrontal cortex (dlPFC). In the performance monitoring region, dorsal anterior cingulate (dACC), activation during error-corrected trials significantly increased with age and reached mature levels in young adulthood. The bootstrap-enhanced elastic net model indicated that person-specific deviations from these adolescent developmental trajectories did not significantly predict QOL in early adulthood, although variable inclusion probabilities suggested that performance monitoring behaviorally and activation in L FEF and R dlPFC may be relatively important predictors.

Conclusions: Findings show that subtle age-related changes in inhibitory control may continue later into adolescence and young adulthood than previously reported. Individual differences in adolescent development of aspects of inhibitory control may potentially be important predictors of QOL in early adulthood, but require further investigation.

Public health significance: Establishing the relationship between individual trajectories of neurocognitive maturation and subsequent QOL may help to inform the development of personal- ized interventions that can be applied during adolescence to promote optimal adult outcomes.