Deviant cognitive control performance is implicated in Attention-Deficit-Hyperactivity-Disorder (ADHD). It is also conjectured to be a potential diagnoser and differentiator between the Inattentive and Hyperactive-Impulsive ADHD types. Reliable measures have not been established due to the variation in published results.

We performed a systematic review and meta-analysis of the literature published up to May 2021 with data on electrophysiological correlates, that is, EEG correlates of cognitive control monitoring (error-related negativity, ERN; error positivity, Pe; correct-response negativity, CRN) in ADHD patients and the efficiency of EEG recordings in differentiating between ADHD types. Multiple databases including PubMed, Scopus, Google Scholar, bioRxiv, and medRxiv were searched for eligible literature. Meta-Analyses were performed through statistical tools provided by the open-source metafor package and separately using the Hedge’s g standardized mean differences.

Meta-Analyses were performed on a shortlisted set of 125 studies involving 7248 participants. To avoid extraneous variables, the sex ratio was maintained at 50:50, and the age groups of participants were equally varied between early teenagers (12-15 years), late teenagers (15-18 years), young adults (21-25), and middle-aged adults (29-37). The ADHD-afflicted group showed reduced ERN (Hedge’s g = −0.58 [CIs: −0.76, −0.35]) and reduced Pe (Hedge’s g = −0.65 [CIs: −0.79, −0.44). The Hyperactive-Impulsive ADHD types (2574/7248 participants) showed an increased CRN (Hedge’s g = 0.68 [CIs: 0.71, 0.29]), while the Inattentive ADHD Types (4674/7248 participants) showed a slightly reduced CRN (Hedge’s g = −0.25 [CIs: −0.31, −0.28]. The prevalence of counted task errors was higher in the teenagers’ group (12-18 years) than the adults’ group (21-37 years).

Results suggest that EEG Pattern Markers (especially Pe and CRN) can act as strong differentiators/diagnosers between the Hyperactive-Impulsive and Inattentive ADHD types. In further development, deep learning classifiers can be built for ADHD types using EEG Markers as Features and statistical values as weights.

Pranjali Awasthi is a 14-year-old researcher working on the overlap of neuroimaging and machine learning at the Neural Dynamics of Control Lab, FIU. She is an avid speaker on topics of AI Awareness and Ethics. Here is a recent feature by the Analytics India Magazine: https://analyticsindiamag.com/how-this-15-year-old-created-a-research-career-in-machine-learning.

New York Institute of Technology MRGA Committee