Estimating individual scores of inattention and impulsivity based on dynamic features of intrinsic connectivity network

Neurosci Lett. 2020 Apr 17:724:134874. doi: 10.1016/j.neulet.2020.134874. Epub 2020 Feb 27.

Abstract

Inattention and impulsivity are the two most important indices for evaluations of ADHD. Currently, inattention and impulsivity were evaluated by clinical scales. The intelligent evaluation of the two indices using machine learning remains largely unexplored. This paper aimed to build regression modes for inattention and impulsivity based on resting state fMRI and additional measures, and discover the associating features for the two indices. To achieve these goals, a cohort of 95 children with ADHD as well as 105 healthy controls were selected from the ADHD-200 database. The raw features were consisted of univariate dynamic estimators of intrinsic connectivity network (ICNs), head motion, and additional measures. The regression models were solved using support vector regression (SVR). The performance of the regression models was evaluated by cross-validations. The performance of regression models based on ICNs outperformed that based on regional measures. The estimated clinical scores were significantly correlated to inattention (r = 0.4 ± 0.02, p < 0.01) and impulsivity (r = 0.31 ± 0.02, p < 0.01). The most associating ICNs are sensorimotor network (SMN) for inattention and executive control network (ECN) for impulsivity. The results suggested that inattention and impulsivity could be estimated using machine learning, and the intra-ICN dynamics could be supplementary features for regression models of clinical scores of ADHD.

Keywords: Impulsivity; Inattention; Intrinsic connectivity network; Regression models; Temporal dynamics.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Attention / physiology
  • Attention Deficit Disorder with Hyperactivity / diagnostic imaging*
  • Attention Deficit Disorder with Hyperactivity / physiopathology
  • Brain / diagnostic imaging*
  • Brain / physiology*
  • Child
  • Cohort Studies
  • Databases, Factual
  • Female
  • Humans
  • Impulsive Behavior / physiology*
  • Logistic Models
  • Magnetic Resonance Imaging / methods
  • Male
  • Nerve Net / diagnostic imaging*
  • Nerve Net / physiology*