Subtyping cognitive profiles in Autism Spectrum Disorder using a Functional Random Forest algorithm

May combine two claims. The quote only partly states this claim.

Chi-square tests found significantly more between-group differences (p < .05) in the cingulo-opercular, visual, and default systems, and differences in inter-system connections in the somato-motor, dorsal attention, and subcortical systems.

“Chi-square tests revealed a significantly greater number of between group differences (p < .05) within the cingulo-opercular, visual, and default systems”

May combine two claims.

Community detection applied to the random forest proximity matrix yielded three ASD and four TD putative subgroups, each with distinct behavioral profiles.

“revealing 3 ASD and 4 TD putative subgroups with unique behavioral profiles.”

May combine two claims.

The random forest model distinguished ASD from TD with 72.7% accuracy, 80.7% specificity, and 63.1% sensitivity.

“Our RF model was 72.7% accurate, with 80.7% specificity and 63.1% sensitivity.”

May combine two claims.

Control subgroup 5 had significantly worse classification accuracy (M=.120, SD=.086) than control subgroups 1, 2, and 4, but not subgroup 3.

“the classification accuracy for subgroup 5 (M = .120 SD=.086) was significantly worse (using a Bonferroni adjusted alpha level of .006 per test) than subgroup 1 [M = .922 SD = .137; t(43)=−13.871, p<.001], subgroup 2 [M = .804 SD = .422; t(9)=−3.910, p=.004], and subgroup 4 [M = .995 SD = .0089; t(7)=−16.903, p<.001], but not subgroup 3 [M = .636 SD = .362; t(9)=−3.411, p=.008].”

May combine two claims.

The third subgroup in both cohorts had the highest spatial span but the lowest accuracy and shortest reaction time on face processing tasks.

“The third subgroup in both cohorts was characterized by highest spatial span, but lowest accuracy and shortest reaction time for the face processing tasks.”

May combine two claims.

Control subgroups 1 and 4 were classified as controls significantly more than chance, while control subgroup 5 was classified as controls significantly less than chance.

“Participants in subgroups 1 [t (38) =19.276, p<.001] and 4 [t (2) =96.00, p<.001] were classified as controls significantly more than chance, while participants in subgroup 5 [t (5) =−10.773, p<.001] were classified as controls significantly less than chance.”

May combine two claims.

ASD subgroups did not differ significantly on the SRS or the ADOS.

“A one-way ANOVA revealed no significant differences between the subgroups on SRS (Figure 2E; F (2, 44) = 0.006, p = 0.994)”

• Also stated in: § Extension of prior Machine Learning studies · plain

  “We found that our subgroups did not differ on the SRS or the ADOS”

• cites [96] Constantino J Gruber C Social responsiveness scale Western Psychological Services 2005Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the control sample, the largest subgroup had 39 individuals; three other subgroups had 5, 5, and 3 individuals; and six controls were not assigned to any community and formed a fifth unspecified subgroup.

“For children without an ASD diagnosis, the largest subgroup comprised 39 individuals; three other subgroups were evenly split with five, five, and three children respectively. Six controls were not identified as part of any community, which were placed into a fifth “unspecified” subgroup.”

May combine two claims.

The second largest subgroup in both cohorts had the smallest spatial span and the highest accuracy and longest reaction times on facial and affect processing tasks.

“The second largest subgroup in both cohorts had the smallest spatial span, and the highest accuracy and longest reaction times for the facial and affect processing tasks.”

May combine two claims.

The chi-squared analysis found significant clustering effects between the cingulo-opercular and default mode systems, the somato-motor hand and default mode systems, the visual and default mode systems, and the subcortical and dorsal attention systems.

“The chi-squared analysis revealed significant clustering effects between the cingulo-opercular system and the default mode system (χ2 = 48.86, p =.0002), the somato-motor hand system and the default mode system (χ2 = 12.81, p =.0016), the visual system and default mode system (χ2 = 11.74, p =.001), and between the subcortical system and the dorsal attention system (χ2 = 35.05, p =.0024).”

May combine two claims.

Community detection identified three putative ASD subgroups and four putative control subgroups.

“The community detection algorithm identified three putative ASD subgroups and four putative control subgroups (Figure 2B).”

May combine two claims.

Control participants showed significant increases in connectivity within the default system and between the dorsal attention and subcortical systems.

“Significant increases in control connectivity were observed within the default system (χ2 = 22.36, p = .0010) and between the dorsal attention system and the subcortical system (χ2 = 11.85, p = .002).”

May combine two claims.

Post-hoc one-way ANOVAs identified 11 significantly different features for control subgroups and 16 for ASD subgroups.

“Post-hoc one-way ANOVAs identified 11 significant different features for control subgroups (F (2, 46) > 3.29, p < 0.0462) and 16 significant different features for ASD subgroups (F (2, 44) > 3.45, p < 0.0405).”

May combine two claims.

Connectivity between the default and somato-motor hand systems showed mixed results, with ASD subgroups deviating from controls.

“Connectivity between the default system and the somato-motor hand system (χ2 = 28.85, p = .0002) showed mixed results, with ASD subgroups deviating from controls.”

May combine two claims.

The largest subgroup in both cohorts performed best on stop and continuous performance tasks and worst on face processing tasks.

“The largest subgroup in both cohorts performed best on stop and continuous performance tasks.”

• Also stated in: § Describe identified subgroups · plain

  “Per sample, the largest subgroups performed best on CPT and stop tasks, and worst on face processing tasks.”

May combine two claims.

In the ASD sample, the largest subgroup had 25 individuals, and the other two subgroups had 13 and 9 individuals.

“For children with an ASD diagnosis, the largest subgroup comprised 25 individuals, while the other two subgroups numbered 13 and 9 children respectively.”

May combine two claims.

ASD subgroups 2 and 3 showed significant increases in connectivity between the cingulo-opercular and default systems and between the dorsal attention and somato-motor hand systems.

“Significant increases in ASD subgroup 2 and 3 between the cingulo-opercular system and the default system (χ2 = 24.01, p =.0002), and between the dorsal attention system and the somato-motor hand system (χ2 = 15.37, p = .0006).”

May combine two claims.

ASD subgroups, control subgroups, and accurately identified subgroups showed significant differences across task measures.

“The ASD subgroups (F(66,1056) = 7.65, p = 7.5*10−54), control subgroups (F(66,1452) = 2.19, p = 2.4*10−7), and accurately identified subgroups (F(33,1716) = 10.64, p = 3.3*10−49) showed significant differences across task”

May combine two claims.

The RF model remained significantly above chance for both ASD (64 percent) and TD (60 percent) individuals.

“Nevertheless, the RF model remained significantly above chance for both ASD (64 percent) and TD (60 percent) individuals.”

May combine two claims.

Inter-system connectivity between the default mode and task control and attention systems (CIP and DOA) is control-like in the first ASD subgroup, as is connectivity between attention and motor systems.

“Inter-system connectivity between the default mode and task control and attention systems (i.e. CIP and DOA) are control-like in the first ASD subgroup, as well as connectivity between attention and motor systems.”

May combine two claims.

Accuracy for the second TD subgroup was almost zero, suggesting they were all being classified with ASD.

“accuracy for the second TD subgroup was almost zero, suggesting that they were all being classified with ASD (Figure S2D)”

May combine two claims.

Gender did not vary in the original analysis between ASD subgroups and between control subgroups.

“Since gender did not vary in the original analysis between ASD subgroups and between control subgroups”

May combine two claims.

The ASD supplemental subgroups varied by age, gender, and IQ.

“The ASD supplemental subgroups varied by age (Figure S5C), gender, and IQ (Figure S6C)”

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

The supplemental RF model identified subgroups that were strongly affected by age, gender, and IQ.

“as the supplemental RF model identified subgroups that were strongly affected by age, gender, and IQ.”

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

Gender differed significantly between the ASD and TD groups (p = 0.025).

“A Fisher’s exact test indicated that gender was significantly different between the two groups (p = 0.025).”

May combine two claims.

Multiple information processing streams may be affected in ASD, but the types of impairment are heterogeneous within the ASD population, with different individuals showing different patterns of difficulty.

“In summary, multiple information processing streams may be affected in individuals with ASD, but the types of impairment may be heterogeneous within the ASD population, with different individuals showing varying patterns of difficulty.”

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

ASD performance on continuous performance tasks is mixed, and individuals with ASD may use an alternative, possibly compensatory, strategy to perform similarly.

“suggesting that individuals with ASD may use an alternative, perhaps compensatory, strategy to perform similarly on CPTs.”

• cites [27] Tye C Attention and inhibition in children with ASD, ADHD and co-morbid ASD + ADHD: an event-related potential study Eur. Child Adolesc. Psychiatry 7 e1210 5 2014 23673307 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

Non-social cognitive mechanisms such as reward, executive function, attention, and visual and auditory processing can influence social behavior presentation independently of specific impairment or domain-specific social cognitive mechanisms.

“A thorough review of cognitive mechanisms underlying ASD suggested that non-social cognitive mechanisms, including reward, executive function, attention, visual and auditory processing, may affect the presentation of social behavior regardless of specific impairment or the existence of domain-specific social cognitive mechanisms7.”

• cites [7] Stone VE Gerrans P What’s domain-specific about theory of mind? Soc Neurosci 1 309 319 2006 18633796 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

The observed functional connectivity differences were largely driven by particular subgroups, which suggests the method may be able to parse variation in brain mechanisms affected by ASD.

“Many of these differences were primarily driven by specific subgroups suggesting that our method could potentially parse the variation in brain mechanisms affected by ASD.”

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

Response inhibition findings in ASD vary despite large samples and similar task designs, and ASD heterogeneity may explain this inconsistency.

“Although sampling variation may explain divergent results, an interesting possibility is that heterogeneity in ASD helps explain the inconsistency across the literature 19.”

• cites [19] Geurts HM van den Bergh SFWM Ruzzano L Prepotent response inhibition and interference control in autism spectrum disorders: Two Meta-Analyses Autism Res 7 407 420 2014 24596300 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Working memory differences between children with and without ASD are inconsistent and may be influenced by sample differences reflecting different ASD subgroups.

“Taken together, all of these findings suggest working memory differences between children with and without ASD are inconsistent, and may be affected by sample differences that comprise different ASD subgroups.”

• cites [14] Chen S-F Deficits in executive functions among youths with autism spectrum disorders: an age-stratified analysis Psychol. Med 1 14 2016 10.1017/S0033291715002238 PMC4873936 26997535Not checked yet

  Not checked in the free run.

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

PDS information was collected once and was not tied to task or MRI visits, limiting inferences from it.

“The PDS information was acquired once for all participants, but was untied to the tasks or MRI visits, which limits our ability to infer from it.”

Classified as probably not a substantial claim.

Structural MRI studies in ASD have reported a wide range of putative biomarkers across independent studies.

“Reviews of structural MRI findings in ASD have found a wide range of putative biomarkers across independent studies 40–42.”

• cites [40] Chen R Jiao Y Herskovits EH Structural MRI in autism spectrum disorder Pediatr Res 69 63R 8R 2011 PMC3081653 21289538 sourceNot checked yet

  Not checked in the free run.

• cites [41] Amaral DG Schumann CM Nordahl CW Neuroanatomy of autism Trends Neurosci 2008 18258309 sourceNot checked yet

  Not checked in the free run.

• cites [42] Brambilla P Brain anatomy and development in autism: Review of structural MRI studies Brain Res. Bull 61 557 569 2003 14519452 sourceNot checked yet

  Not checked in the free run.

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

Individuals with ASD have difficulty searching for the eye region on a face.

“Individuals with ASD show impairments in searching for the eye region on a face 31.”

• cites [31] Pruett JR Impaired Eye Region Search Accuracy in Children with Autistic Spectrum Disorders PLoS One 8 2013 PMC3597629 23516446 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Within the ASD population, symptoms, severity, and adaptive behavior impairment vary.

“The variability in symptoms, severity, and adaptive behavior impairment within the ASD population 4 complicates the development of effective treatments and improved diagnostic measures.”

• cites [4] Hill EL Executive dysfunction in autism Trends Cogn. Sci 8 26 32 2004 14697400 sourceNot checked yet

  Not checked in the free run.

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

Facial emotion recognition results in ASD are mixed.

“However, results on facial emotion recognition are more mixed 36.”

• cites [36] Annaz D Karmiloff-Smith A Johnson MH Thomas MSC A cross-syndrome study of the development of holistic face recognition in children with autism, Down syndrome, and Williams syndrome J. Exp. Child Psychol 102 456 486 2009 19193384Not checked yet

  Not checked in the free run.

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

An earlier study sub-grouped ASD participants by symptom severity, verbal IQ, and age, which increased classification rates by as much as 10%.

“In an earlier study, ASD participants were sub-grouped on the basis of symptom severity, verbal IQ, and age, which caused classification rates to increase by as much as 10% 65.”

• cites [65] Katuwal GJ Baum SA Cahill ND Michael AM Divide and Conquer?: Sub-Grouping of ASD Improves ASD Detection Based on Brain Morphometry PLoS One 11 1 24 2016 PMC4827874 27065101 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Liu et al. used leave-one-out cross validation, which improves classification accuracy within a test dataset but may reduce generalizability to other datasets.

“Liu et al used a leave-one-outcross validation (LOOCV) strategy, which improves classification accuracy within a test dataset, but may reduce the generalizability of the model to other datasets.”

May combine two claims.

The subgroup connectivity differences overlapped substantially with ASD-control differences, suggesting that normal variation in mechanisms also affected by ASD may cause variation in how ASD manifests.

“These differences overlapped substantially with the differences observed between ASD and controls (Figure 5A), suggesting that normal variation in mechanisms that are also affected by ASD may cause variation in how ASD may manifest 1,91.”

• cites [1] Constantino JN Charman T Diagnosis of autism spectrum disorder: reconciling the syndrome, its diverse origins, and variation in expression Lancet. Neurol 15 279 291 2016 26497771 sourceNot checked yet

  Not checked in the free run.

• cites [91] Szatmari P Developmental trajectories of symptom severity and adaptive functioning in an inception cohort of preschool children with autism spectrum disorder JAMA psychiatry 72 276 83 2015 25629657 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Higher performance has been reported for behavior when constructing a model using visual face scanning (88.1%) or goal-oriented reach (96.7%) measures.

“Higher performance has been reported for behavior when constructing a model using visual face scanning (88.1%; 92) or goal-oriented reach (96.7%; 93) measures.”

• cites [92] Liu W Li M Yi L Identifying children with autism spectrum disorder based on their face processing abnormality: A machine learning framework Autism Res 9 888 898 2016 27037971 sourceNot checked yet

  Not checked in the free run.

• cites [93] Crippa A Use of Machine Learning to Identify Children with Autism and Their Motor Abnormalities J. Autism Dev. Disord 45 2146 2156 2015 25652603 sourceNot checked yet

  Not checked in the free run.

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

These functional connectivity findings are consistent with prior reports of resting-state functional connectivity MRI differences between typically developing and ASD samples.

“These findings are consistent with prior reports of rsfcMRI differences between TD and ASD samples (see: Discussion).”

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

Wang et al. used whole-brain functional connectivity to correctly predict ASD with 83% accuracy in a sample of 58 participants.

“Wang et al., using whole-brain functional connectivity, correctly predicted ASD with 83% accuracy in a sample of 58 participants 63.”

• cites [63] Wang H Chen C Fushing H Extracting Multiscale Pattern Information of fMRI Based Functional Brain Connectivity with Application on Classification of Autism Spectrum Disorders PLoS One 7 1 14 2012 PMC3466274 23056205 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

The fact that control subgroups also showed misclassification suggests that variation in such skills may represent broad cognitive subgroups independent of diagnosis, whose variation may affect ASD symptom presentation.

“On the other hand, the fact that control subgroups also showed misclassification suggests that variation in such skills may represent the existence of broad cognitive subgroups that are independent of diagnosis, whose variation may impact the presentation of ASD symptoms 1.”

• cites [1] Constantino JN Charman T Diagnosis of autism spectrum disorder: reconciling the syndrome, its diverse origins, and variation in expression Lancet. Neurol 15 279 291 2016 26497771Not checked yet

  Not checked in the free run.

May combine two claims.

Non-predictive features did not contribute meaningfully to the classification model and did not materially affect results.

“Despite this liberal inclusion, these non-predictive features did not contribute meaningfully to the classification model and thus did not affect results materially (supplementary materials).”

• cites [73] Breiman LEO Random Forests Mach. Learn 45 5 32 2001Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Recent classification studies incorporating brain measures have shown comparable results to the initial classification and suggest that heterogeneity of clinically relevant ASD subgroups may limit high classification accuracy.

“Recent classification studies incorporating brain measures have shown comparable results to our initial classification and further suggest that heterogeneity of clinically relevant ASD subgroups may limit high classification accuracy.”

Classified as probably not a substantial claim.

These subgroup connectivity findings should be interpreted cautiously because the data are not predictive of diagnosis.

“These findings should be interpreted cautiously, however, because these data are not predictive of diagnosis.”

May combine two claims.

Crippa et al. also used leave-one-out cross validation and were limited in sample size.

“Crippa et al. also used a LOOCV validation strategy, and were also limited in sample size.”

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

Delay Discounting data that failed three pre-specified validity criteria were treated as missing in analyses.

“Data that did not meet validity criteria were treated as missing in analyses.”

• cites [76] Johnson MW Bickel WK An algorithm for identifying nonsystematic delay-discounting data Exp. Clin. Psychopharmacol 16 264 2008 18540786 PMC2765051 sourceNot checked yet

  Not checked in the free run.

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

Murdaugh et al. used intra-DEF connectivity to predict ASD with 96% accuracy in a sample of 27 participants.

“Murdaugh et al. used the intra-DEF connectivity to predict ASD with 96% accuracy in a sample of 27 participants 62.”

• cites [62] Murdaugh DL Differential Deactivation during Mentalizing and Classification of Autism Based on Default Mode Network Connectivity PLoS One 7 2012 PMC3501481 23185536Not checked yet

  Not checked in the free run.

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

Jamal et al. used EEG activity during task switching to predict ASD with 95% accuracy in a sample of 24 participants.

“Jamal et al. used EEG activity during task switching to predict ASD with 95% accuracy in a sample of 24 participants 64.”

• cites [64] Jamal W Classification of autism spectrum disorder using supervised learning of brain connectivity measures extracted from synchrostates J. Neural Eng 11 046019 2014 24981017 sourceNot checked yet

  Not checked in the free run.

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

Duchesnay et al. found that PET imaging could predict ASD with 88% accuracy in a sample of 26 participants.

“Duchesnay et al. found that PET imaging could be used to predict ASD with 88% accuracy in a sample of 26 participants 61.”

• cites [61] Duchesnay E Feature selection classification of imbalanced datasets. Application to PET images of children with autistic spectrum disorders Neuroimage 57 1003 1014 2011 21600290 sourceNot checked yet

  Not checked in the free run.

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

Our data highlight the importance of considering heterogeneity for such tests.

“Our data highlights the importance of considering heterogeneity for such tests.”

• AssumptionRests on an assumption: The cognitive tasks used in the random forest model measure domains affected by ASD and are relevant to ASD classification.

May combine two claims.

Machine learning approaches using imaging data have shown that validation accuracy decreases as sample size increases, suggesting that small sample sizes may overfit the data.

“Machine learning approaches using imaging data have shown that validation accuracy decreases as the sample size increases, suggesting that these small sample sizes may be overfitting the data 65,68.”

• cites [65] Katuwal GJ Baum SA Cahill ND Michael AM Divide and Conquer?: Sub-Grouping of ASD Improves ASD Detection Based on Brain Morphometry PLoS One 11 1 24 2016 PMC4827874 27065101 sourceNot checked yet

  Not checked in the free run.

• cites [68] Sabuncu MR Konukoglu E Clinical Prediction from Structural Brain MRI Scans?: A Large-Scale Empirical Study Neuroinformatics 2014 10.1007/s12021-014-9238-1 PMC4303550 25048627 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

ASD subgroups had similar autism severity but varied in other ways.

“suggesting that ASD subgroups had similar autism severity but varied in other ways.”

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

Children with ASD have shown altered visual system responses to stacks of oriented lines.

“Children with an ASD have shown altered visual system responses to stacks of oriented lines 99”

• cites [99] Vandenbroucke MW Scholte HS van Engeland H Lamme VA Kemner C A neural substrate for atypical low-level visual processing in autism spectrum disorder Brain 131 1013 1024 2008 18192288 sourceNot checked yet

  Not checked in the free run.

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

The second ASD subgroup may have both altered visual and attention mechanisms.

“and the second ASD subgroup may have both”

May combine two claims.

Taken together, the findings suggest that clinical heterogeneity may emerge from normal variation in cognitive profiles, consistent with a recent study showing that clinical heterogeneity within ASD may be driven by normative development.

“Taken together, these findings suggest that clinical heterogeneity may emerge from normal variation in cognitive profiles, and are consistent with a recent study showing that clinical heterogeneity within ASD may be driven by normative development97.”

• cites [97] van der Meer JMJ Homogeneous Combinations of ASD--ADHD Traits and Their Cognitive and Behavioral Correlates in a Population-Based Sample J. Atten. Disord 2014 1087054714533194 24819924Not checked yet

  Not checked in the free run.

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

Literature suggests that biases may occur when group differences might differ by controlling variables, but the features important for classification are not associated with those variables.

“There is some literature101 that suggests such biases may occur when the differences in groups might differ by the controlling variables, but the features important for classification (i.e. in this case the behavioral measures) are not associated with those variables (i.e., here, age or gender).”

• cites [101] Miller GA Chapman JP Misunderstanding analysis of covariance J. Abnorm. Psychol 110 40 48 2001 11261398 sourceNot checked yet

  Not checked in the free run.

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

The inconsistency in findings may be interpreted as further evidence of heterogeneity within ASD.

“The inconsistency in findings may be interpreted as further evidence of heterogeneity within ASD.”

May combine two claims.

Because no differences in symptom severity between subgroups were observed, the findings are more likely to reflect typical variation in neural mechanisms underlying cognitive performance rather than manifestations of ASD symptoms.

“Because we did not observe differences in symptom severity between subgroups, the findings above are more likely to reflect typical variation in neural mechanisms underlying cognitive performance, as opposed to manifestations of ASD symptoms.”

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

Differences in diagnostic criteria may explain the apparent contradiction.

“However, differences in diagnostic criteria may explain some of the apparent contradiction here.”

May combine two claims.

The fMRI data were not used in the random forest model, so they served as an independent validation set.

“Since this data was never used in the RF model”

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

The ANOVA chi-squared analysis may be underpowered and is not definitive.

“We would like to emphasize that the ANOVA chi-squared analysis may be underpowered 88 and, though enticing, is not definitive.”

• cites [88] Eggebrecht AT Joint Attention and Brain Functional Connectivity in Infants and Toddlers Cereb. Cortex 1 12 2017 10.1093/cercor/bhw403 PMC5452276 28062515 sourceNot checked yet

  Not checked in the free run.

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

The first ASD subgroup may have altered visual processing mechanisms.

“hint that the first ASD subgroup may have altered visual processing mechanisms”

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

Variation within the ASD subgroups may be nested within the normal variation found in brain networks across typical children.

“it is possible that variation within the ASD subgroups identified here may actually be “nested” within the normal variation found in brain networks across typical children 1,71,91”

• cites [1] Constantino JN Charman T Diagnosis of autism spectrum disorder: reconciling the syndrome, its diverse origins, and variation in expression Lancet. Neurol 15 279 291 2016 26497771 sourceNot checked yet

  Not checked in the free run.

• cites [71] Fair Da Distinct neural signatures detected for ADHD subtypes after controlling for micro-movements in resting state functional connectivity MRI data Front. Syst. Neurosci 6 80 2012 23382713 PMC3563110 sourceNot checked yet

  Not checked in the free run.

• cites [91] Szatmari P Developmental trajectories of symptom severity and adaptive functioning in an inception cohort of preschool children with autism spectrum disorder JAMA psychiatry 72 276 83 2015 25629657 sourceNot checked yet

  Not checked in the free run.

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

Children with ASD at rest have exhibited altered default mode (DEF) functional connectivity.

“at rest they’ve exhibited altered DEF functional connectivity 52”

• cites [52] Monk CS Abnormalities of intrinsic functional connectivity in autism spectrum disorders Neuroimage 47 764 772 2009 19409498 PMC2731579Not checked yet

  Not checked in the free run.

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

The third ASD subgroup may have altered attention mechanisms.

“the third ASD subgroup may have altered attention mechanisms”

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

Prior work suggests that cognitive subtypes within ASD may be similar to cognitive subtypes found in typical populations.

“In addition, prior work suggests that cognitive subtypes within ASD may be similar to cognitive subtypes found in typical populations98.”

• cites [98] Rommelse NNJ van der Meer JMJ Hartman CA Buitelaar JK Cognitive Profiling Useful for Unraveling Cross-Disorder Mechanisms Support for a Step-Function Endophenotype Model Clin. Psychol. Sci 4 957 970 2016Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

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

Dorsal attention network organization has been shown to be altered in ASD.

“as well as the dorsal attention network organization 51”

• cites [51] Ray S Structural and functional connectivity of the human brain in autism spectrum disorders and attention-deficit/hyperactivity disorder: A rich club-organization study Hum. Brain Mapp 2014 10.1002/hbm.22603 PMC4319550 25116862 sourceNot checked yet

  Not checked in the free run.

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

Differences between default mode and cingulo-opercular systems may be consistent with altered rich-club organization.

“while differences between DEF and cingulo-opercular systems may be consistent with altered rich-club organization 51”

• cites [51] Ray S Structural and functional connectivity of the human brain in autism spectrum disorders and attention-deficit/hyperactivity disorder: A rich club-organization study Hum. Brain Mapp 2014 10.1002/hbm.22603 PMC4319550 25116862 sourceNot checked yet

  Not checked in the free run.

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

Similar concerns have been found when using parametric tests like ANCOVA in psychiatric research.

“Similar concerns have been found when using parametric tests like analysis of covariance (ANCOVA) in psychiatric research101.”

• cites [101] Miller GA Chapman JP Misunderstanding analysis of covariance J. Abnorm. Psychol 110 40 48 2001 11261398 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

The finding that subgroup differences were similar in both ASD and TD samples may appear inconsistent with prior studies showing an effect of ASD on the relationship between cognitive measures and task performance.

“Our finding that the differences between subgroups were similar in both ASD and TD samples may appear inconsistent with prior studies that show an effect of ASD on the relationship between cognitive measures and task performance12,27,37,38.”

• cites [12] Joseph RM Steele SD Meyer E Tager-Flusberg H Self-ordered pointing in children with autism: Failure to use verbal mediation in the service of working memory? Neuropsychologia 43 1400 1411 2005 15989932 sourceNot checked yet

  Not checked in the free run.

• cites [27] Tye C Attention and inhibition in children with ASD, ADHD and co-morbid ASD + ADHD: an event-related potential study Eur. Child Adolesc. Psychiatry 7 e1210 5 2014 23673307 sourceNot checked yet

  Not checked in the free run.

• cites [37] Barton JJS Are patients with social developmental disorders prosopagnosic? Perceptual heterogeneity in the Asperger and socio-emotional processing disorders Brain 127 1706 1716 2004 15215211 sourceNot checked yet

  Not checked in the free run.

• cites [38] Hefter RL Manoach DS Barton JJS Perception of facial expression and facial identity in subjects with social developmental disorders Neurology 65 1620 1625 2005 16301491 sourceNot checked yet

  Not checked in the free run.

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

Such an interpretation is consistent with the findings from the original RF.

“Such an interpretation is consistent with the findings from the original RF.”

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

Such tasks would involve visual processing.

“Such tasks would involve visual processing”

May combine two claims.

The study extends prior findings to ASD and establishes a predictive model, which provides some clinical validity to the identified subgroups.

“Our study here extends the prior findings to ASD and establishes a predictive model, which provides some clinical validity to the identified subgroups.”

• AssumptionRests on an assumption: The cognitive tasks used in the random forest model measure domains affected by ASD and are relevant to ASD classification.

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

Subcortical-cortical connectivity has been shown to be altered in ASD.

“sub-cortical cortical connectivity has been shown to be altered 55,57”

• cites [55] Di Martino A The autism brain imaging data exchange: towards a large-scale evaluation of the intrinsic brain architecture in autism Mol. Psychiatry 19 659 67 2014 23774715 PMC4162310 sourceNot checked yet

  Not checked in the free run.

• cites [57] Cerliani L Increased Functional Connectivity Between Subcortical and Cortical Resting-State Networks in Autism Spectrum Disorder JAMA Psychiatry 72 1 11 2015 PMC5008437 26061743 sourceNot checked yet

  Not checked in the free run.

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

Prior work by Fair et al. has shown similar heterogeneity in both typically developing and ADHD children, and accounting for this heterogeneity improved diagnostic accuracy.

“Prior work by Fair et al. has shown similar heterogeneity in both TD and ADHD children; as with Katuwal, taking into account this heterogeneity improved diagnostic accuracy 71.”

• cites [71] Fair Da Distinct neural signatures detected for ADHD subtypes after controlling for micro-movements in resting state functional connectivity MRI data Front. Syst. Neurosci 6 80 2012 23382713 PMC3563110 sourceNot checked yet

  Not checked in the free run.

May combine two claims. The quote only partly states this claim.

Altered connections between subcortical and dorsal attention networks are consistent with altered subcortical-cortical connectivity and altered dorsal attention network organization.

“which is consistent with altered connections between subcortical and dorsal attention networks”

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

The subgroup chi-squared ANOVAs hint that the identified subgroups may reflect differences in both mechanisms relevant to an ASD diagnosis and mechanisms that reflect variation across the subgroups.

“Nevertheless, the subgroup chi-squared ANOVAs hint that the identified subgroups may reflect differences in both mechanisms relevant to an ASD diagnosis, and mechanisms that reflect variation across the subgroups.”

May combine two claims.

Because the authors are confident in the ASD diagnosis, they suspect that variation among the three subgroups reflects typical variation in cognitive mechanisms that may be independent of autism symptom severity but influence ASD presentation.

“Because we are confident in the ASD diagnosis, we suspect that the variation between these three subgroups reflects typical variation in cognitive mechanisms, which may be independent of autism symptom severity but influence ASD presentation14,91.”

• cites [14] Chen S-F Deficits in executive functions among youths with autism spectrum disorders: an age-stratified analysis Psychol. Med 1 14 2016 10.1017/S0033291715002238 PMC4873936 26997535 sourceNot checked yet

  Not checked in the free run.

• cites [91] Szatmari P Developmental trajectories of symptom severity and adaptive functioning in an inception cohort of preschool children with autism spectrum disorder JAMA psychiatry 72 276 83 2015 25629657 sourceNot checked yet

  Not checked in the free run.

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

Such variation is consistent with the literature and may reflect typical heterogeneity variability related to the presentation of ASD.

“such variation is consistent with the literature and may reflect typical heterogeneity variability related to the presentation of ASD.”

May combine two claims.

The study used a team of experts to confirm ASD diagnosis per individual, whereas prior studies often used only a DSM diagnosis plus one or two instruments assessing autism symptom severity.

“Our study used a team of experts to confirm ASD diagnosis per individual, whereas these prior studies often used only a DSM diagnosis plus one or two instruments that assess autism symptom severity (e.g. the ADOS and/or ADIR).”

May combine two claims.

Variation in facial task performance may be implicated in some children with autism but not others.

“Variation in facial task performance may be implicated in some children with autism 39, but not others 37.”

• cites [39] Tanaka JW The perception and identification of facial emotions in individuals with Autism Spectrum Disorders using the Let’s Face It! Emotion Skills Battery J. Child Psychol. Psychiatry 53 1259 1267 2012 22780332 PMC3505257 sourceNot checked yet

  Not checked in the free run.

• cites [37] Barton JJS Are patients with social developmental disorders prosopagnosic? Perceptual heterogeneity in the Asperger and socio-emotional processing disorders Brain 127 1706 1716 2004 15215211 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

The rs-fcMRI analysis provides some evidence that speaks to the validity of the subgroups but does not fully validate them.

“Therefore, while the rs-fcMRI analysis provides some evidence that speaks to the validity of the subgroups, the analysis does not fully validate them.”

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

Tying the distance measure directly to diagnosis increases confidence that identified subgroups are relevant to the clinical question.

“By tying the distance measure directly to diagnosis we can be more confident that the identified subgroups are relevant to the clinical question.”

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

The proximity method employed by the RF approach implicitly handles missing data.

“Second, the proximity method employed by the RF approach implicitly handles missing data.”

May combine two claims.

The study had sufficient power to examine the effect of ASD on resting state data but was underpowered when examining differences between subgroups.

“While we had sufficient power to examine the effect of ASD on resting state data 88, we were underpowered when examining differences between subgroups.”

• cites [88] Eggebrecht AT Joint Attention and Brain Functional Connectivity in Infants and Toddlers Cereb. Cortex 1 12 2017 10.1093/cercor/bhw403 PMC5452276 28062515 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Taken together, the findings suggest that the original RF was driven by variation in typical cognitive profiles, whereas the supplemental RF may be affected by variation in gender and age.

“Taken together, the findings suggest that the original RF was driven by variation in typical cognitive profiles, whereas the supplemental RF may be affected by variation in gender and age.”

Classified as probably not a substantial claim.

The identified subgroups in this sample are putative.

“However, we want to emphasize that the identified subgroups in this sample are putative.”