Abstract:Attention-deficit/hyperactivity disorder (ADHD) is a highly prevalent and heterogeneous neurodevelopmental disorder, which is diagnosed using subjective symptom reports. Machine learning classifiers have been utilized to assist in the development of neuroimaging-based biomarkers for objective diagnosis of ADHD. However, existing basic model-based studies in ADHD report suboptimal classification performances and inconclusive results, mainly due to the limited flexibility for each type of basic classifier to appropriately handle multi-dimensional source features with varying properties. This study applied ensemble learning techniques (ELTs), a meta-algorithm that combine several basic machine learning models into one predictive model in order to decrease variance, bias, or improve predictions, in multimodal neuroimaging data collected from 72 young adults, including 36 probands (18 remitters and 18 persisters of childhood ADHD) and 36 group-matched controls. All currently available optimization strategies for ELTs (i.e., voting, bagging, boosting and stacking techniques) were tested in a pool of semifinal classification results generated by seven basic classifiers. The high-dimensional neuroimaging features for classification included regional cortical gray matter (GM) thickness and surface area, GM volume of subcortical structures, volume and fractional anisotropy of major white matter fiber tracts, pair-wise regional connectivity and global/nodal topological properties of the functional brain network for cue-evoked attention process. As a result, the bagging-based ELT with the base model of support vector machine achieved the best results, with significant improvement of the area under the receiver of operating characteristic curve (0.89 for ADHD vs. controls and 0.9 for ADHD persisters vs. remitters). Features of nodal efficiency in right inferior frontal gyrus, right middle frontal (MFG)-inferior parietal (IPL) functional connectivity, and right amygdala volume significantly contributed to accurate discrimination between ADHD probands and controls; higher nodal efficiency of right MFG greatly contributed to inattentive and hyperactive/impulsive symptom remission, while higher right MFG-IPL functional connectivity strongly linked to symptom persistence in adults with childhood ADHD. Considering their improved robustness than the commonly implemented basic classifiers, findings suggest that ELTs may have the potential to identify more reliable neurobiological markers for neurodevelopmental disorders.

Beyond out-of-sample: robust and generalizable multivariate neuroanatomical patterns of psychiatric problems in youth

Limited generalizability of multivariate brain-based dimensions of child psychiatric symptoms

Brain Functional Connectome Defines a Transdiagnostic Dimension Shared by Cognitive Function and Psychopathology in Preadolescents

Generalizable prediction of childhood ADHD symptoms from neurocognitive testing and youth characteristics

Evaluation of behavioral variance/covariance explained by the neuroimaging data through a pattern‐based regression

Voxel-wise multivariate analysis of brain-psychosocial associations in adolescents reveals six latent dimensions of cognition and psychopathology

Brain-phenotype predictions can survive across diverse real-world data

Multivariate brain-behaviour associations in psychiatric disorders

Multimethod investigation of the neurobiological basis of ADHD symptomatology in children aged 9-10: baseline data from the ABCD study

Multitrait analysis to decipher the intertwined genetic architecture of neuroanatomical phenotypes and psychiatric disorders.

Big Connectome Imaging Data in Psychiatric Disorders

Longitudinally stable, brain-based predictive models mediate the relationships between childhood cognition and socio-demographic, psychological and genetic factors

A shared neural basis underlying psychiatric comorbidity

Mapping common grey matter volume deviation across child and adolescent psychiatric disorders

Individualized functional connectome identified generalizable biomarkers for psychiatric symptoms in transdiagnostic patients

Multimodal neural correlates of childhood psychopathology

Morphometric dis-similarity between cortical and subcortical areas underlies cognitive function and psychiatric symptomatology: a preadolescence study from ABCD

Investigating cross-sectional and longitudinal relationships between brain structure and distinct dimensions of externalizing psychopathology in the ABCD sample

Evaluating the Neuroimaging-Genetic Prediction of Symptom Changes in Individuals with ADHD

Multimodal neuroimaging-based prediction of adult outcomes in childhood-onset ADHD using ensemble learning techniques