CoChat
Sign Up
Cited by 107Functional Brain Connectivity StudiesMental Health Research TopicsAttention Deficit Hyperactivity Disorder

Organizing Heterogeneous Samples Using Community Detection of GIMME-Derived Resting State Functional Networks

Open access
Gates, Kathleen M., Molenaar, Peter C. M., Iyer, Swathi P., Nigg, Joel T., Fair, Damien A.·2014·PLoS ONE
DOI 10.1371/journal.pone.0091322PMC PMC3958357
Open this paper in CoChat — free accountOpen PDF

Abstract

Clinical investigations of many neuropsychiatric disorders rely on the assumption that diagnostic categories and typical control samples each have within-group homogeneity. However, research using human neuroimaging has revealed that much heterogeneity exists across individuals in both clinical and control samples. This reality necessitates that researchers identify and organize the potentially varied patterns of brain physiology. We introduce an analytical approach for arriving at subgroups of individuals based entirely on their brain physiology. The method begins with Group Iterative Multiple Model Estimation (GIMME) to assess individual directed functional connectivity maps. GIMME is one of the only methods to date that can recover both the direction and presence of directed functional connectivity maps in heterogeneous data, making it an ideal place to start since it addresses the problem of heterogeneity. Individuals are then grouped based on similarities in their connectivity patterns using a modularity approach for community detection. Monte Carlo simulations demonstrate that using GIMME in combination with the modularity algorithm works exceptionally well--on average over 97% of simulated individuals are placed in the accurate subgroup with no prior information on functional architecture or group identity. Having demonstrated reliability, we examine resting-state data of fronto-parietal regions drawn from a sample (N = 80) of typically developing and attention-deficit/hyperactivity disorder (ADHD) -diagnosed children. Here, we find 5 subgroups. Two subgroups were predominantly comprised of ADHD, suggesting that more than one biological marker exists that can be used to identify children with ADHD based from their brain physiology. Empirical evidence presented here supports notions that heterogeneity exists in brain physiology within ADHD and control samples. This type of information gained from the approach presented here can assist in better characterizing patients in terms of outcomes, optimal treatment strategies, potential gene-environment interactions, and the use of biological phenomenon to assist with mental health.

AI snapshot

Written from the paper’s full text.

TL;DR

The paper introduces a data-driven method that combines GIMME-derived individual directed functional connectivity maps with Newman modularity community detection to identify subgroups of individuals based on brain physiology alone. In simulations it recovered subgroups with high accuracy, and in resting-state fMRI from 80 children it found five subgroups, with two predominantly ADHD, supporting heterogeneity within ADHD and control samples.

Key findings

  • In Monte Carlo simulations (100 individuals, 10 brain regions), GIMME recovered group-level connections for all 100 datasets and, on average, 99.73% (SD = 0.09%, range = 95–100%) of individual connections accurately in presence and direction.
  • Community detection placed on average 97.23% (SD = 1.87%, range = 92–100%) of simulated individuals into the correct subgroup without prior group information.
  • In the empirical sample of 80 children (32 ADHD, 48 typically developing controls), the method found 5 subgroups; ADHD and control participants were distributed across subgroups at greater than chance levels (χ² = 12.32, df = 4, p = .015), while sex distribution did not differ from chance (χ² = 3.65, df = 4, p = .456).
  • Comparing ADHD and control as homogeneous diagnostic groups revealed no significant differences in path weights, but subgroup analyses showed different connectivity patterns: e.g., subgroup B had weaker L dlPFC→L IPS connectivity, while subgroup D had stronger L dlPFC→L IPS connectivity and decreased L FC→R FC connectivity.

Methods

  • Resting-state fMRI data were analyzed from simulated datasets and an empirical sample of 80 children aged 7–12; 11 fronto-parietal ROIs were selected, and individual directed functional connectivity maps were estimated with GIMME within a unified structural equation modeling framework.
  • Each individual’s contemporaneous connection weights were vectorized, correlated with every other individual’s vector to form a similarity matrix, binarized at a data-driven threshold, and submitted to Newman’s modularity community detection algorithm.
  • Threshold selection used stability of the modularity index Q across 100 runs and a reachability criterion; robustness of community structure was tested via perturbation and variation of information, and subgroup connection weights were compared using ANOVA with FDR correction at alpha = 0.05.

Limitations

  • The paper states that community detection algorithms have proliferated without formal comparison, that Newman’s algorithm is widely used but likely not optimal for all situations, and that many fMRI community detection approaches were tested on larger adjacency matrices and may not be optimal for smaller matrices where nodes are participants.
  • The paper also states that similarity-matrix construction could vary with preprocessing choices such as global signal regression, that lagged paths could alternatively be used, and that individual-level paths were omitted because their heterogeneity produced groups too small to be useful; future work is needed for non-normally distributed data.
  • Evident from the text: the empirical findings come from a modest cross-sectional sample (N = 80) with no independent replication or outcome/treatment validation, and the “protective” versus “risk” subgroup interpretations are proposed as hypotheses rather than confirmed clinical predictions.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1371/journal.pone.0091322

Cites

58 of 58 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 57 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Using GIMME in combination with the modularity algorithm places on average over 97% of simulated individuals in the accurate subgroup with no prior information on functional architecture or group identity.

    “Monte Carlo simulations demonstrate that using GIMME in combination with the modularity algorithm works exceptionally well - on average over 97% of simulated individuals are placed in the accurate subgroup with no prior information on functional architecture or group identity.”
    • Also stated in: § Simulated data · boosted
      “The community detection algorithm described above then identified the subgroups excellently: 97.23% (std. = 1.87%, range = 92–100%) of the individuals within each data set were placed in subgroups with those who shared their simulated brain map pattern.”
    Quote in the paperSubstantial claim
  • Findingown data

    In a sample of 80 typically developing and ADHD-diagnosed children, resting-state fronto-parietal data yielded 5 subgroups.

    “Having demonstrated reliability, we examine resting-state data of fronto-parietal regions drawn from a sample (N = 80) of typically developing and attention-deficit/hyperactivity disorder (ADHD) -diagnosed children. Here, we find 5 subgroups.”
    Quote in the paperSubstantial claim
  • Findingown data

    Two of the 5 subgroups were predominantly comprised of ADHD-diagnosed children.

    “Two subgroups were predominantly comprised of ADHD, suggesting that more than one biological marker exists that can be used to identify children with ADHD based from their brain physiology.”
    Quote in the paperSubstantial claim
  • Findingown data

    Subgroup D had decreased connectivity from L FC to R FC when compared to the average of the other groups.

    “Indeed, subgroup D (risk) appears to have decreased connectivity from the L FC to the R FC when compared to the average of the other groups”
    Quote in the paperSubstantial claim
  • Findingreasoning

    The Q index becomes unstable at certain thresholds.

    “As depicted in Figure 2 and the top panel of Figure 3, the Q index becomes unstable at certain thresholds.”
    Quote in the paperSubstantial claim
  • May combine two claims. 30
  • The quote only partly states this claim. 1
  • Classified as probably not a substantial claim. 30

  • May combine two claims.

    Findingown data

    The subgroups were consistently recovered in the simulated data and were robust to minor perturbations.

    “The subgroups were consistently recovered in the simulated data and were robust to minor perturbations.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingcitation

    Subgroup B had lower directed connectivity strength from L dlPFC to L IPS, consistent with previous research on deficits in attention.

    “Importantly, subgroup B had lower directed connectivity strength from the left dorsolateral prefrontal cortex (L dlPFC) to the left intraparietal sulcus (L IPS), which is consistent with previous research on deficits in attention [44].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [44] C Liston, 2009, Psychosocial stress reversibly disrupts prefrontal processing and attentional control, Proceedings of the National Academy of Sciences of the United States of America source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    Subgroups B and D contain a high percentage of the ADHD children, who make up a large part of these subgroups.

    “Subgroups B and D contain a high percentage of the ADHD children, who make up a large part of these subgroups.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The smaller subgroups C and E are disproportionately control, but contain relatively small percentages of the control sample.

    “The smaller subgroups C and E are disproportionately control, but contain relatively small percentages of the control sample.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The community assignments remain similar for simulated individuals until a perturbation level of about .5, at which point 20% of the vertices are assigned different communities than the original and variation of information increases steadily past this point.

    “The community assignments remain similar for the simulated individuals until a perturbation level of about .5 (i.e., random assignment of 50% of the edges), at which point 20% of the vertices are assigned different communities than the original and we see this variation of information increasing steadily past this point.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Findingown data

    Subgroup B had strong directed influence of L FC on R FC, like subgroup A.

    “but like subgroup A had strong directed influence of the left frontal cortex (L FC) on the R FC.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The connectivity map in Figure 5b provides the connectivity pattern of the fronto-parietal region set common for the majority of individuals in the whole sample (the group map).

    “The connectivity map depicted in Figure 5b provides the connectivity pattern of the fronto-parietal region set common for the majority of individuals in the whole sample (i.e. connections found to be significant for the majority, or the “group map”).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    The curves for the empirical and random data remain distinct and thus the network of individuals can be said to show community structure.

    “Regarding the robustness of the solution as quantified using VI, the curves for the empirical and random data remain distinct and thus the network of individuals can be said to show community structure (see Figure 6b; [24]).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [24] B Karrer, 2008, Robustness of community structure in networks, Physical Review E source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    The other ADHD-dominant subgroup had increased connectivity between these two regions but had decreased connectivity between the right and left frontal cortex.

    “However, the other ADHD-dominant subgroup had increased connectivity between these two regions but had decreased connectivity between the right and left frontal cortex.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The high degree of heterogeneity with regards to the individual-level paths necessitated that we omit these when creating the similarity matrices since including individual-level results via a distance measure (as opposed to correlation) resulted in groups that were too small to offer any value over individual-level analysis.

    “Additionally, the high degree of heterogeneity with regards to the individual-level paths necessitated that we omit these when creating the similarity matrices since including individual-level results via a distance measure (as opposed to correlation) resulted in groups that were too small to offer any value over individual-level analysis.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The majority of ADHD-diagnosed children fell into two subgroups, about a third were spread across the three subgroups comprised of predominantly typically developing control.

    “Although the majority of ADHD-diagnosed children fell into two subgroups, about a third were spread across the three subgroups comprised of predominantly typically developing control.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Premisecitation

    Brain processes can best be thought of as coordinated activity of disparate regions across time.

    “Two, given that brain processes can best be thought of as coordinated activity of disparate regions across time [14], [15], we necessitated an approach for arriving at precise brain connectivity maps that quantifies relations among brain regions across time for each individual.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [14] O Sporns, 2004, Organization, development and function of complex brain networks, Trends Cogn Sci source

      Not checked in the free run.

    • cites [15] G Tononi, 1994, A measure for brain complexity: relating functional segregation and integration in the nervous system, Proceedings of the National Academy of Sciences source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Assuming within-group homogeneity will likely cause misleading results at the aggregate level and miss important features for subpopulations within a given category.

    “Moreover, assuming within-group homogeneity will likely cause misleading results at the aggregate level [8]–[10] in addition to missing important features for subpopulations within a given category.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [8]–[10]

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    The assumptions that neurobiology for predefined diagnostic groups is distinct from other groups and that individuals are homogeneous within predefined groups are rarely met.

    “A critical review of literature reveals that these assumptions are rarely met [5]–[7].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [5]–[7]

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    Brain connectivity maps have been used to identify systematic differences between subgroups and conditions in both clinical and non-clinical populations.

    “Brain connectivity maps have been used to identify systematic differences between subgroups and conditions in both clinical and non-clinical populations (e.g., [5], [13]).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [5] FG Hillary, 2011, Examining working memory task acquisition in a disrupted neural network, Brain source

      Not checked in the free run.

    • cites [13] AM Beltz, 2013, Changes in alcohol-related brain networks across the first year of college: a prospective pilot study using fMRI effective connectivity mapping, Addict Behav source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Most evidence for this phenomenon has been from neuropsychological data and behavior or symptom report, with recent evidence suggesting heterogeneity in the biological components as well.

    “Most evidence for this phenomenon has been from neuropsychological data and behavior or symptom report (e.g., [2]), with recent evidence suggesting heterogeneity in the biological components as well [48].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [2] DA Fair, 2012, Distinct neuropsychological subgroups in typically developing youth inform heterogeneity in children with ADHD, Proc Natl Acad Sci U S A source

      Not checked in the free run.

    • cites [48] D Fair, 2013, Distinct Neural Signatures Detected for ADHD Subtypes After Controlling for Micro-Movements in Resting State Functional Connectivity MRI Data, Frontiers in Systems Neuroscience source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    There appears to be some connections that exist for the majority of individuals while there also exists a large degree of heterogeneity in the structure of connections as evidenced by the extra connections uncovered at the individual level.

    “Hence there appears to be some connections that exist for the majority of individuals while there also exists a large degree of heterogeneity in the structure of connections as evidenced by the extra connections uncovered at the individual level.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The preprocessing steps are thought to reduce spurious variance unlikely to reflect neuronal activity.

    “These preprocessing steps are thought to reduce spurious variance unlikely to reflect neuronal activity [38].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [38] MD Fox, 2007, Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging, Nat Rev Neurosci source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationreasoning

    The development of treatments, preventative, and intervention strategies for a given individual also suffer from this confound.

    “The development of treatments, preventative, and intervention strategies for a given individual also suffer from this confound.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Studies examining the etiology of any given disorder typically ignore the potential heterogeneity that exists within current symptom-based classifications.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Variation in connection weights of specific connections between two ROIs has long been a focal point of fMRI research.

    “Variation in the connection weights of specific connections between two given ROIs has long been a focal point of fMRI research [17].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [17] J Kim, 2007, Unified structural equation modeling approach for the analysis of multisubject, multivariate functional MRI data, Hum Brain Mapp source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Subgroups A, C, and E are subgroups with low likelihood for ADHD and as such have an organization of brain physiology that we term "protective."

    “Taken together, subgroups A, C, and E are subgroups with low likelihood for ADHD and as such have an organization of brain physiology that we term “protective.””
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Heterogeneity exists within both control and ADHD populations.

    “These findings demonstrate that heterogeneity exists within both control and ADHD populations”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisenone given

    Small literatures have made attempts to parse fMRI data across individuals using various statistical methods, but none have been sufficiently accepted to fall into wide use.

    “Although small literatures have made attempts to parse fMRI data across individuals using various statistical methods (e.g., [11]), none have been sufficiently accepted to fall into wide use.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [11] F Kherif, 2003, Group analysis in functional neuroimaging: selecting subjects using similarity measures, NeuroImage source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationown data

    The empirical data example offered concrete evidence for theories that have hitherto been untested.

    “The empirical data example offered concrete evidence for theories that have hitherto been untested.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Unified SEM contains both lagged and contemporaneous directed relations among regions, making it ideal for biological systems that likely contain these effects.

    “Unified SEM contains both lagged and contemporaneous directed relations among regions, making it ideal for biological systems that likely contain these effects as detailed in [18].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [17] J Kim, 2007, Unified structural equation modeling approach for the analysis of multisubject, multivariate functional MRI data, Hum Brain Mapp source

      Not checked in the free run.

    • cites [18] KM Gates, 2010, Automatic search for fMRI connectivity mapping: an alternative to Granger causality testing using formal equivalences among SEM path modeling, VAR, and unified SEM, Neuroimage source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The method is robust and easily applied, utilizing functional connectivity networks to arrive at brain-based subgroups with no prior classification information.

    “Taken together, our approach is a robust and easily applied method that utilizes functional connectivity networks to arrive at brain-based subgroups with no prior classification information.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The degree of similarity to expect and variability in connection weights has been informed by previous research that used individual-level connectivity maps to examine a priori defined subgroups.

    “The degree of similarity to expect as well as the variability in connection weights has been informed by previous research that used individual-level connectivity maps to examine a priori defined subgroups (e.g., [5])”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [5] FG Hillary, 2011, Examining working memory task acquisition in a disrupted neural network, Brain source

      Not checked in the free run.

  • May combine two claims.

    Study design and conductown data

    Psychiatric diagnoses were assigned with adequate agreement (kappa>.75 for all disorders with base rate >5% in a larger sample, kappa >.80 for ADHD).

    “They independently assigned all appropriate diagnoses, with adequate agreement (kappa>.75 for all disorders with base rate >5% in a larger sample of several hundred children they have reviewed) and kappa >.80 for ADHD.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Researchers have made strong assertions that clinical diagnoses may result from multiple etiologies.

    “In particular, researchers have made strong assertions that clinical diagnoses may result from multiple etiologies [3], [4].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [3] FR Volkmar, 2004, Autism and pervasive mental disorders, Journal of Child Psychology and Psychiatry

      Not checked in the free run.

    • cites [4] Gottesman II, 2003, The endophenotype concept in psychiatry: Etymology and strategic intentions, American Journal of Psychiatry source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    The 11 fronto-parietal regions were selected based on their implied role in adaptive task level control that surfaces in resting state studies and atypical nature in studies of ADHD.

    “These regions termed the fronto-parietal network (see Figure 5a) were selected based on their implied role in adaptive task level control that surfaces in resting state studies [40], [41] and atypical nature in studies of ADHD [42], [43].”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [40] NU Dosenbach, 2007, Distinct brain networks for adaptive and stable task control in humans, Proc Natl Acad Sci U S A source

      Not checked in the free run.

    • cites [41] NU Dosenbach, 2008, A dual-networks architecture of top-down control, Trends Cogn Sci source

      Not checked in the free run.

    • cites [42]–[43]

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    This possibility has been noted in ADHD as well as Autism Spectrum Disorder.

    “This possibility has been noted in ADHD as well as Autism Spectrum Disorder [3], [47].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [3] FR Volkmar, 2004, Autism and pervasive mental disorders, Journal of Child Psychology and Psychiatry

      Not checked in the free run.

    • cites [47] C Betancur, 2011, Etiological heterogeneity in autism spectrum disorders: more than 100 genetic and genomic disorders and still counting, Brain Res source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    There appears to be different physiological markers within the ADHD diagnostic category as well as different profiles within the control sample.

    “Taken together, there appears to be different physiological markers within the ADHD diagnostic category as well as different profiles within the control sample.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Within-group heterogeneity washed out effects of interest that we were able to isolate by enabling for heterogeneity within the diagnostic categories.

    “Hence it appears the within-group heterogeneity washed out effects of interest that we were able to isolate by enabling for heterogeneity within the diagnostic categories.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    The heterogeneity presents a significant barrier in the development of genetic and biological markers to assist in accurate diagnoses for a given disorder.

    “However, it does present a significant barrier in the development of genetic and biological markers to assist in accurate diagnoses for a given disorder.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Studies examining the etiology of any given disorder typically ignore the potential heterogeneity that exists within current symptom-based classifications.

  • May combine two claims.

    Interpretationown data

    Subgroups B and D appear to have brain organizations that place them at higher likelihood for having ADHD.

    “Subgroups B and D appear to have brain organizations that place them at higher likelihood for having ADHD.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    GIMME reliably recovered the presence and true direction of connections at the group and individual levels and is one of the few approaches that can do so.

    “As seen previously [10], GIMME reliably recovered the presence and true direction of connections at the group and individual levels and is one of the few approaches that can do so [46].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [10] KM Gates, 2012, Group search algorithm recovers effective connectivity maps for individuals in homogeneous and heterogeneous samples, Neuroimage source

      Not checked in the free run.

    • cites [46] JA Mumford, 2014, Bayesian networks for fMRI: A primer, Neuroimage source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    The method presented here is a step towards the goal of identifying subgroups that are most meaningful with regard to clinical translation.

    “The method presented here is a step towards this goal.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisereasoning

    The proliferation of community detection programs has occurred quickly enough such that they have yet to be formally compared and evaluated to identify which ones are best for which situations.

    “The proliferation of community detection programs has occurred quickly enough such that they have yet to be formally compared and evaluated to identify which ones are best for which situations (e.g., number of individuals in each “true” group, number of subgroups in a sample, unequal group sizes, small sample sizes) and what scientific questions.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Data in the ADHD literature examining functional differences in the diagnostic subtypes (i.e. Combined type, and Inattentive type) lend support for this particular finding.

    “Data in the ADHD literature examining functional differences in the diagnostic subtypes (i.e. Combined type, and Inattentive type) lend support for this particular finding [2].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [2] DA Fair, 2012, Distinct neuropsychological subgroups in typically developing youth inform heterogeneity in children with ADHD, Proc Natl Acad Sci U S A source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    Conducting individual-level analysis would result in unique maps from which no meaningful inferences could be made.

    “whereas conducting individual-level analysis would result in unique maps from which no meaningful inferences could be made.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Premisenone given

    Data of a substantial sample size are often capable of being split into a multitude of valid subgroup arrangements.

    “Data of a substantial sample size are often capable of being split into a multitude of valid subgroup arrangements.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    We used GSR because it offers a number of benefits such as improving the correspondence between resting-state correlations and anatomy and motion correction at the time series level.

    “We used GSR because it offers a number of benefits such as improving the correspondence between resting-state correlations and anatomy [54] and motion correction at the time series level [55], [56].”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [54] MD Fox, 2009, The global signal and observed anticorrelated resting state brain networks, Journal of neurophysiology source

      Not checked in the free run.

    • cites [55] TD Satterthwaite, 2013, Heterogeneous impact of motion on fundamental patterns of developmental changes in functional connectivity during youth, Neuroimage source

      Not checked in the free run.

    • cites [56] CG Yan, 2013, A comprehensive assessment of regional variation in the impact of head micromovements on functional connectomics, Neuroimage source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    There is the possibility that GSR altered connectivity patterns, which in turn may change the degree of similarity of individuals and thus the resulting community designations.

    “There is the possibility that GSR altered connectivity patterns [57], [58], which in turn may change the degree of similarity of individuals and thus the resulting community designations.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [57] SJ Gotts, 2013, The perils of global signal regression for group comparisons: a case study of Autism Spectrum Disorders, Frontiers in human neuroscience source

      Not checked in the free run.

    • cites [58] ZS Saad, 2012, Trouble at rest: how correlation patterns and group differences become distorted after global signal regression, Brain connectivity source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Grouping all ADHD individuals together may not capture all of the informative characteristics of the disorder and in this sample, washed out meaningful differences.

    “Hence grouping all ADHD individuals together may not capture all of the informative characteristics of the disorder and in this sample, washed out meaningful differences.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    The connectivity mapping results may vary based on preprocessing decisions, such as global signal regression (GSR).

    “For instance, the connectivity mapping results may vary based on preprocessing decisions, such as global signal regression (GSR).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    A long line of research in the social sciences has urged researchers to examine processes at the individual level rather than aggregate as is typically done in cross-sectional studies.

    “A long line of research in the social sciences has urged researchers to examine processes at the individual level rather than aggregate as is typically done in cross-sectional studies [45], [49]–[51].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [45] PC Molenaar, 2004, A manifesto on psychology as idiographic science: Bringing the person back into scientific psychology, this time forever, Measurement

      Not checked in the free run.

    • cites [49]–[51]

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    The work supports the potential of a biologically based nosology in the future.

    “Nonetheless, the work supports the potential of a biologically based nosology in the future.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Premisenone given

    Sample-level paths are normally distributed across individuals and thus appropriate for obtaining correlation values denoting similarity in values for use with the community detection algorithm.

    “We utilize sample-level paths since they are normally distributed across individuals and thus appropriate for obtaining correlation values denoting similarity in values for use with the community detection algorithm.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    Many community detection approaches in the fMRI literature have been tested on relatively large adjacency matrices and may not be optimal for smaller matrices where the nodes are participants.

    “Importantly, many community detection approaches in the fMRI literature have been tested on relatively large adjacency matrices and, while useful, may not be optimal for smaller matrices where the nodes are participants.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    What determines a given demarcation rests predominantly on the features chosen to show similarities or differences between individuals and the algorithm chosen to demarcate the sample.

    “What determines a given demarcation rests predominantly on the features (in our case connections) chosen to show similarities or differences between individuals and the algorithm chosen to demarcate the sample.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Typically developing controls may be heterogeneous in functional brain architecture, or brain "profiles", is not a new concept.

    “Importantly, the fact that typically developing controls may be heterogeneous in functional brain architecture, or brain “profiles”, is not a new concept either.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Premisereasoning

    The Newman algorithm, as applied here, is widely used, but is likely not optimal for all situations.

    “The Newman algorithm [19], as applied here, is widely used, but is likely not optimal for all situations.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [19] ME Newman, 2006, Modularity and community structure in networks, Proc Natl Acad Sci U S A source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Amidst the heterogeneity within diagnostic categories, subgroups emerged that were predominantly clinical or control, supporting some degree of biological validity to the existing nosology.

    “Amidst the heterogeneity within diagnostic categories, subgroups emerged that were predominantly clinical or control, supporting some degree of biological validity to the existing nosology.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    Contemporaneous paths seem to hold information regarding neuronal relationships.

    “Since contemporaneous paths seem to hold information regarding neuronal relationships, we chose to use these.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    There has been an influx of algorithms that seek to improve upon the traditionally used community detection ones.

    “Along with this growth there has been an influx of algorithms that seek to improve upon the traditionally used ones (e.g., [53]).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [53] VD Blondel, 2008, Fast unfolding of communities in large networks, Journal of Statistical Mechanics: Theory and Experiment source

      Not checked in the free run.

55 of 55

  • The authors state this as an assumption.

    Clinical investigations of many neuropsychiatric disorders rely on the assumption that diagnostic categories and typical control samples each have within-group homogeneity.

    “Clinical investigations of many neuropsychiatric disorders rely on the assumption that diagnostic categories and typical control samples each have within-group homogeneity.”

  • The paper relies on this without support.

    Studies examining the etiology of any given disorder typically ignore the potential heterogeneity that exists within current symptom-based classifications.

    “While it is largely agreed that the above possibility exists, studies examining the etiology of any given disorder typically ignore the potential heterogeneity that exists within current symptom-based classifications.”

  • The paper relies on this without support.

    Analyses are conducted as though both the diagnostic group and the comparison control group represent two homogeneous populations.

    “In accordance with the current standard, analyses are conducted as though both the diagnostic group and the comparison control group represent two homogeneous populations.”

  • The paper relies on this without support.

    rs-fcMRI is developmentally and contextually more variable than brain structural measures.

    “A benefit of rs-fcMRI is that it is developmentally and contextually more variable than brain structural measures, making it a highly attractive place to start towards the goal of identifying subgroups of individuals based on similar brain processes.”

  • The paper relies on this without support.

    rs-fcMRI can be administered across any age range, species, or cognitive ability.

    “Furthermore, it can be administered at across any age range, species, or cognitive ability.”

  • The paper relies on this without support.

    Prior rs-fcMRI analyses have revealed ongoing information processing between spatially disparate regions of the brain even during rest, and differences in these processes relate to cognitive performance and psychiatric disorders.

    “Prior rs-fcMRI analyses have revealed that there is ongoing information processing between spatially disparate regions of the brain even during rest, and differences in these observed processes relate to cognitive performance as well as psychiatric disorders.”

  • The paper relies on this without support.

    Sample-level connections will have normally distributed estimates across individuals.

    “An added benefit is that sample-level connections will have normally distributed estimates across individuals, making it appropriate for the correlation analysis to follow.”

  • The paper relies on this without support.

    Researchers commonly select thresholds that provide the highest modularity index.

    “Researchers commonly select thresholds (or solutions) that provide the highest modularity index.”

  • The paper relies on this without support.

    The ADHD and TDC diagnostic categories are valid and meaningful groupings for the analyses.

    “we placed and analyzed the individuals into groups based on their diagnostic status.”

  • The paper relies on this without support.

    The sample of children is representative of the broader ADHD and typically developing populations to which the findings are generalized.

    “These findings demonstrate that heterogeneity exists within both control and ADHD populations”
10 of 10

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Male percentage conflicts with the reported sex totals

    Major

    Table 3 lists 32 children with ADHD and 48 controls. The reported male percentages imply approximately 25 male children with ADHD and 22 male controls, or 47 males overall, not the 51 stated for the sample.

  • Simulated-data standard deviation is incompatible with its range

    Minor

    Across 100 data sets with a mean of 99.73%, a value as low as 95% alone requires a standard deviation of at least about 0.48 percentage points. The reported 0.09% is therefore incompatible with the reported range.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. FR Volkmar, 2004, Autism and pervasive mental disorders, Journal of Child Psychology and Psychiatry, 45: 135, 170

Cited by

Sorted by citations

Related

Authors
Gates, K. M., Molenaar, P. C. M., Iyer, S. P., Nigg, J. T., & Fair, D. A.
Affiliations
  • University of North Carolina at Chapel Hill (US)
  • Pennsylvania State University (US)
  • Oregon Health & Science University (US)
Corresponding author
Gates, Kathleen M.
ORCID
Venue
PLoS ONE
Publisher
Public Library of Science (PLoS)
Published
2014-03-18
Volume
9
Issue
3
Pages
e91322
ISSN
1932-6203 (ISSN-L)
DOI
10.1371/journal.pone.0091322
PMC
PMC3958357
OpenAlex
W2047219509
Cited by
107
References
58
Citations by year
2026: 4 · 2025: 2 · 2024: 5 · 2023: 6 · 2022: 12 · 2021: 6 · 2020: 14 · 2019: 17 · 2018: 9 · 2017: 14 · 2016: 9 · 2015: 6 · 2014: 3
FWCI
5.95
Type
journal article
Language
en
Open access
Licence
Locations
  • PLoS ONE (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Carolina Digital Repository (University of North Carolina at Chapel Hill) (submittedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • UNC Libraries
Funders
  • National Science Foundation (1157220)
  • National Institutes of Health (1157220, K99-R00, K99/R00 MH091238, MH091238)
  • Oregon Clinical and Translational Research Institute
  • National Institute of Mental Health (K99/R00 MH091238, K99MH091238, R01MH115357, R01 MH86654, R00MH091238, R01MH096773, R01MH086654)
  • National Institute of Biomedical Imaging and Bioengineering (R21EB015573)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Functional Brain Connectivity Studies
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Mental Health Research Topics
  • Health Sciences › Medicine › Psychiatry and Mental health › Attention Deficit Hyperactivity Disorder
Keywords
group iterative multiple model estimation, directed functional connectivity, resting-state networks, community detection, modularity, patient heterogeneity, subgroup identification, attention deficit hyperactivity disorder, children, frontoparietal connectivity, biomarkers, neuropsychiatric disorders
MeSH terms
Attention Deficit Disorder with Hyperactivity, Brain Mapping, Child, Computer Simulation, Female, Frontal Lobe, Humans, Magnetic Resonance Imaging, Male, Neural Pathways, Parietal Lobe, Residence Characteristics, Models, Statistical, Biomarkers, Case-Control Studies, Genetic Heterogeneity, Gene-Environment Interaction

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-06.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-06.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review