Meeting Banner
Abstract #1175

Motor Cortex Functional Connectivity Signatures of Autism

Mary Beth Nebel1, 2, Ani Eloyan3, Anita Barber, 12, Brian S. Caffo4, James J. Pekar, 12, Stewart H. Mostofsky1, 2

1Kennedy Krieger Institute, Baltimore, MD, United States; 2Johns Hopkins School of Medicine, Baltimore, MD, United States; 3Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; 4Johns Hopkins University, Baltimore, MD, United States

Children with autism spectrum disorders (ASD) struggle with a host of motor behaviors, which may reflect abnormal connectivity within motor control and learning networks. Our objective was to estimate how well functional connectivity (FC) among subregions of the motor cortex (M1) discriminate individuals with ASD from neurotypical (NT) participants using a large, heterogeneous resting state fMRI dataset (368 ASD and 412 NT). Using a multinomial logistic regression model with demographic factors and M1 correlations as predictors and disease status as the outcome, we identified FC signatures of ASD that are consistent with deficits in complex multi-joint coordination associated with ASD.

Keywords

abide abnormal acquisition adjusted among analyses anatomical assessed autism beth bioengineering boosted brain calculation characteristics children clinical collected computed computing connectivity consistency contributing control core correlation correlations cortex courses deficits demographic derived developing diagnosis discriminate disease disorders displacement disruptions enable foundation full functional funding gender general generalized global grants hand health heterogeneous highest impairments included independent individual influence institute intelligence inter johns joint kernel learning least limb limited linear logistic mainly males many medicine mental model motion motor multinomial national near networks neurological noisy normalization normalized normally nuisance organization outcome pairs parcel participant parts performance poses potentially power powerful predicting prediction predictive predictors preliminary prevalent previously prior public quantitative quotient reduce reflect registered regressed regression relate related resting retest sample scalable scale school scripts segmentation several signatures sites slice sources space spatial speaks spectrum status stroke subject subjects subregions system template third tool typically unified variability variables volume white whose years