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

Distributed associations among white matter hyperintensities and structural brain networks with fluid cognition in healthy aging

White matter hyperintensities (WMHs) are associated with age-related cognitive impairment and increased risk of Alzheimer's disease. However, the manner by which WMHs contribute to cognitive impairment is unclear. Using a combination of predictive modeling and network neuroscience, we investigated the relationship between structural white matter connectivity and age, fluid cognition, and WMHs in 68 healthy adults (18-78 years). Consistent with previous work, WMHs were increased in older adults and exhibited a strong negative association with fluid cognition. Extending previous work, using predictive modeling, we demonstrated that age, WMHs, and fluid cognition were jointly associated with widespread alterations in structural connectivity. Subcortical-cortical connections between the thalamus/basal ganglia and frontal and parietal regions of the default mode and frontoparietal networks were most prominent. At the network level, both age and WMHs were negatively associated with network density and communicability, and positively associated with modularity. Spatially, WMHs were most prominent in arterial zones served by the middle cerebral artery and associated lenticulostriate branches that supply subcortical regions. Finally, WMHs overlapped with all major white matter tracts, most prominently in tracts that facilitate subcortical-cortical communication and are implicated in fluid cognition, including the anterior thalamic-radiations and forceps minor. Finally, results of mediation analyses suggest that whole-brain WMH load influences age-related decline in fluid cognition. Thus, across multiple levels of analysis, we showed that WMHs were increased in older adults and associated with altered structural white matter connectivity and network topology involving subcortical-cortical pathways critical for fluid cognition.

Decomposition of phenotypic heterogeneity in autism reveals distinct and coherent genetic programs

Unraveling the phenotypic and genetic complexity of autism is extremely challenging yet critical for understanding the biology, inheritance, trajectory, and clinical manifestations of the many forms of the condition. Here, we leveraged broad phenotypic data from a large cohort with matched genetics to characterize classes of autism and their patterns of core, associated, and co-occurring traits, ultimately demonstrating that phenotypic patterns are associated with distinct genetic and molecular programs. We used a generative mixture modeling approach to identify robust, clinically-relevant classes of autism which we validate and replicate in a large independent cohort. We link the phenotypic findings to distinct patterns of de novo and inherited variation which emerge from the deconvolution of these genetic signals, and demonstrate that class-specific common variant scores strongly align with clinical outcomes. We further provide insights into the distinct biological pathways and processes disrupted by the sets of mutations in each class. Remarkably, we discover class-specific differences in the developmental timing of genes that are dysregulated, and these temporal patterns correspond to clinical milestone and outcome differences between the classes. These analyses embrace the phenotypic complexity of children with autism, unraveling genetic and molecular programs underlying their heterogeneity and suggesting specific biological dysregulation patterns and mechanistic hypotheses.

Participation as a Predictor of Quality of Life among Japanese Children with Neurodevelopmental Disorders Analyzed Using a Machine Learning Algorithm

Participation is important for children's quality of life (QOL). This study aimed to identify participation factors that influence QOL among Japanese children with neurodevelopmental disorders. Ninety-two Japanese parents of children with neurodevelopmental disorders participated in this study. The parents completed the parent version of the Kid- and Kiddo-KINDL health-related QOL questionnaire and the Participation and Environment Measure for Children and Youth. The data were examined using the random forest algorithm to analyze the participation factors that affected the children's QOL. The analyses revealed that school and community environmental factors that affected participation were the most important predictors of QOL among children. As school and community environments can significantly impact the QOL of children with neurodevelopmental disorders, greater focus should be placed on participation in environmental contexts.

Understanding the subtypes of non-suicidal self-injury: A new conceptual framework based on a systematic review

Non-suicidal self-injury (NSSI) is a significant public health problem, but there is no consistent evidence of its risk factors. One possibility is that there are subtypes NSSI that have different risk factors and clinical symptoms. In this review we evaluated the evidence of subtypes to determine if there were consistent subtypes of NSSI that emerged across studies. Four databases (Medline; Embase; PsycINFO; Web of Science) were searched to identify studies that used data-driven approaches and were published before November 9, 2022. There were 21 studies with 23 unique samples for review. Most of the included studies used NSSI symptoms or personal characteristics as the subtyping indicators, revealing 2-5 subtypes of NSSI. Variations in subtyping indicators, sample characteristics, and statistical methods may have contributed to the inconsistent number and characteristics of subtypes across studies. A new conceptual framework is proposed to integrate these diverse findings, highlighting the important roles of NSSI function and psychological pain in differentiating NSSI subtypes. This framework sheds light on the differences among self-injurers and offers insights for future endeavors to address the complexities of NSSI.

Comprehension of irony in autistic children: The role of theory of mind and executive function

Although previous studies have examined irony comprehension in autistic children and potential impact factors, the relationship between theory of mind (ToM), executive function (EF), symptoms of autism, and comprehension of irony in this population remains largely unknown. This study explored irony comprehension in autistic children and examined the roles of ToM and EF in linking autism symptoms to deficits in irony comprehension. Twenty autistic children were compared with 25 typically developing (TD) children in an irony story picture task, ToM task, and EF task. The results showed that autistic children had impaired comprehension of irony compared with TD children, and performance on ironic stories showed a significant moderate discriminatory effect in predicting autistic children. A ToM deficit has also been proposed for autistic children. Comprehension of irony was significantly correlated with second-order ToM (2nd ToM) but was not significantly correlated with any components of EF. Moreover, 2nd ToM can predict the level of irony comprehension and mediate the relationship between symptoms of autism and irony comprehension. Taken together, these findings suggest that irony comprehension may offer a potential cognitive marker for quantifying syndrome manifestations in autistic children, and 2nd ToM may provide insight into the theoretical mechanism underlying the deficit in irony comprehension in this population.

Novel machine learning approaches for improving the reproducibility and reliability of functional and effective connectivity from functional MRI

Objective.New measures of human brain connectivity are needed to address gaps in the existing measures and facilitate the study of brain function, cognitive capacity, and identify early markers of human disease. Traditional approaches to measure functional connectivity (FC) between pairs of brain regions in functional MRI, such as correlation and partial correlation, fail to capture nonlinear aspects in the regional associations. We propose a new machine learning based measure of FC (ML.FC) which efficiently captures linear and nonlinear aspects.Approach.To capture directed information flow between brain regions, effective connectivity (EC) metrics, including dynamic causal modeling and structural equation modeling have been used. However, these methods are impractical to compute across the many regions of the whole brain. Therefore, we propose two new EC measures. The first, a machine learning based measure of effective connectivity (ML.EC), measures nonlinear aspects across the entire brain. The second, Structurally Projected Granger Causality (SP.GC) adapts Granger Causal connectivity to efficiently characterize and regularize the whole brain EC connectome to respect underlying biological structural connectivity. The proposed measures are compared to traditional measures in terms ofreproducibilityand theability to predict individual traitsin order to demonstrate these measures' internal validity. We use four repeat scans of the same individuals from the Human Connectome Project and measure the ability of the measures to predict individual subject physiologic and cognitive traits.Main results.The proposed new FC measure ofML.FCattains high reproducibility (mean intra-subjectR2of 0.44), while the proposed EC measure ofSP.GCattains the highest predictive power (meanR2across prediction tasks of 0.66).Significance.The proposed methods are highly suitable for achieving high reproducibility and predictiveness and demonstrate their strong potential for future neuroimaging studies.

Maternal Perinatal Stress Trajectories and Negative Affect and Amygdala Development in Offspring

OBJECTIVE

Maternal psychological stress during pregnancy is a common risk factor for psychiatric disorders in offspring, but little is known about how heterogeneity of stress trajectories during pregnancy affect brain systems and behavioral phenotypes in infancy. This study was designed to address this gap in knowledge.

METHODS

Maternal anxiety, stress, and depression were assessed at multiple time points during pregnancy in two independent low-risk mother-infant cohorts (N=115 and N=2,156). Trajectories in maternal stress levels in relation to infant negative affect were examined in both cohorts. Neonatal amygdala resting-state functional connectivity MRI was examined in a subset of one cohort (N=60) to explore the potential relationship between maternal stress trajectories and brain systems in infants relevant to negative affect.

RESULTS

Four distinct trajectory clusters, characterized by changing patterns of stress over time, and two magnitude clusters, characterized by severity of stress, were identified in the original mother-infant cohort (N=115). The magnitude clusters were not associated with infant outcomes. The trajectory characterized by increasing stress in late pregnancy was associated with blunted development of infant negative affect. This relationship was replicated in the second, larger cohort (N=2,156). In addition, the trajectories that included increasing or peak maternal stress in late pregnancy were related to stronger neonatal amygdala functional connectivity to the anterior insula and the ventromedial prefrontal cortex in the exploratory analysis.

CONCLUSIONS

The trajectory of maternal stress appears to be important for offspring brain and behavioral development. Understanding heterogeneity in trajectories of maternal stress and their influence on infant brain and behavioral development is critical to developing targeted interventions.

Probing the overarching continuum theory: data-driven phenotypic clustering of children with ASD or ADHD

The clinical validity of the distinction between ADHD and ASD is a longstanding discussion. Recent advances in the realm of data-driven analytic techniques now enable us to formally investigate theories aiming to explain the frequent co-occurrence of these neurodevelopmental conditions. In this study, we probe different theoretical positions by means of a pre-registered integrative approach of novel classification, subgrouping, and taxometric techniques in a representative sample (N = 434), and replicate the results in an independent sample (N = 219) of children (ADHD, ASD, and typically developing) aged 7-14 years. First, Random Forest Classification could predict diagnostic groups based on questionnaire data with limited accuracy-suggesting some remaining overlap in behavioral symptoms between them. Second, community detection identified four distinct groups, but none of them showed a symptom profile clearly related to either ADHD or ASD in neither the original sample nor the replication sample. Third, taxometric analyses showed evidence for a categorical distinction between ASD and typically developing children, a dimensional characterization of the difference between ADHD and typically developing children, and mixed results for the distinction between the diagnostic groups. We present a novel framework of cutting-edge statistical techniques which represent recent advances in both the models and the data used for research in psychiatric nosology. Our results suggest that ASD and ADHD cannot be unambiguously characterized as either two separate clinical entities or opposite ends of a spectrum, and highlight the need to study ADHD and ASD traits in tandem.

School distress and the school attendance crisis: a story dominated by neurodivergence and unmet need

Background

The Covid-19 pandemic has brought into sharp focus a school attendance crisis in many countries, although this likely pre-dates the pandemic. Children and young people (CYP) struggling to attend school often display extreme emotional distress before/during/after school. We term this School Distress. Here we sought to elucidate the characteristics of the CYP struggling to attend school in the United Kingdom.

Methods

Using a case-control, concurrent embedded mixed-method research design, 947 parents of CYP with experience of School Distress completed a bespoke online questionnaire (February/March 2022), alongside an age-matched control group (n = 149) and a smaller group of parents who electively home-educate (n = 25).

Results

In 94.3% of cases, school attendance problems were underpinned by significant emotional distress, with often harrowing accounts of this distress provided by parents. While the mean age of the CYP in this sample was 11.6 years (StDev 3.1 years), their School Distress was evident to parents from a much younger age (7.9 years). Notably, 92.1% of CYP currently experiencing School Distress were described as neurodivergent (ND) and 83.4% as autistic. The Odds Ratio of autistic CYP experiencing School Distress was 46.61 [95% CI (24.67, 88.07)]. Autistic CYP displayed School Distress at a significantly earlier age, and it was significantly more enduring. Multi-modal sensory processing difficulties and ADHD (among other neurodivergent conditions) were also commonly associated with School Distress; with School Distress CYP having an average of 3.62 NDs (StDev 2.68). In addition, clinically significant anxiety symptomology (92.5%) and elevated demand avoidance were also pervasive. Mental health difficulties in the absence of a neurodivergent profile were, however, relatively rare (6.17%). Concerningly, despite the striking levels of emotional distress and disability reported by parents, parents also reported a dearth of meaningful support for their CYP at school.

Conclusion

While not a story of exclusivity relating solely to autism, School Distress is a story dominated by complex neurodivergence and a seemingly systemic failure to meet the needs of these CYP. Given the disproportionate number of disabled CYP impacted, we ask whether the United Kingdom is upholding its responsibility to ensure the "right to an education" for all CYP (Human Rights Act 1998).

An overview of clustering methods with guidelines for application in mental health research

Cluster analyzes have been widely used in mental health research to decompose inter-individual heterogeneity by identifying more homogeneous subgroups of individuals. However, despite advances in new algorithms and increasing popularity, there is little guidance on model choice, analytical framework and reporting requirements. In this paper, we aimed to address this gap by introducing the philosophy, design, advantages/disadvantages and implementation of major algorithms that are particularly relevant in mental health research. Extensions of basic models, such as kernel methods, deep learning, semi-supervised clustering, and clustering ensembles are subsequently introduced. How to choose algorithms to address common issues as well as methods for pre-clustering data processing, clustering evaluation and validation are then discussed. Importantly, we also provide general guidance on clustering workflow and reporting requirements. To facilitate the implementation of different algorithms, we provide information on R functions and libraries.

A graph theoretic approach to neurodegeneration: five data-driven neuropsychological subtypes in mild cognitive impairment

Mild cognitive Impairment (MCI) is notoriously heterogenous in terms of clinical presentation, neuroimaging correlates, and subsequent progression. Predicting who will progress to dementia, which type of dementia, and over what timeframe is challenging. Previous work has attempted to identify MCI subtypes using neuropsychological measures in an effort to address this challenge; however, there is no consensus on approach, which may account for some of the variability. Using a hierarchical community detection approach, we examined cognitive subtypes within an MCI sample (from the Alzheimer's Disease Neuroimaging Initiative [ADNI] study). We then examined whether these subtypes were related to biomarkers (e.g., cortical volumes, fluorodeoxyglucose (FDG)-positron emission tomography (PET) hypometabolism) or clinical progression. We identified five communities (i.e., cognitive subtypes) within the MCI sample: 1) predominantly memory impairment, 2) predominantly language impairment, 3) cognitively normal, 4) multidomain, with notable executive dysfunction, 5) multidomain, with notable processing speed impairment. Community membership was significantly associated with 1) cortical volume in the hippocampus, entorhinal cortex, and fusiform cortex; 2) FDG PET hypometabolism in the posterior cingulate, angular gyrus, and inferior/middle temporal gyrus; and 3) conversion to dementia at follow up. Overall, community detection as an approach appears a viable method for identifying unique cognitive subtypes in a neurodegenerative sample that were linked to several meaningful biomarkers and modestly with progression at one year follow up.

Different levels of visuospatial abilities linked to differential brain correlates underlying visual mental segmentation processes in autism

The neural underpinnings of enhanced locally oriented visual processing that are specific to autistics with a Wechsler's Block Design (BD) peak are largely unknown. Here, we investigated the brain correlates underlying visual segmentation associated with the well-established autistic superior visuospatial abilities in distinct subgroups using functional magnetic resonance imaging. This study included 31 male autistic adults (15 with (AUTp) and 16 without (AUTnp) a BD peak) and 28 male adults with typical development (TYP). Participants completed a computerized adapted BD task with models having low and high perceptual cohesiveness (PC). Despite similar behavioral performances, AUTp and AUTnp showed generally higher occipital activation compared with TYP participants. Compared with both AUTnp and TYP participants, the AUTp group showed enhanced task-related functional connectivity within posterior visuoperceptual regions and decreased functional connectivity between frontal and occipital-temporal regions. A diminished modulation in frontal and parietal regions in response to increased PC was also found in AUTp participants, suggesting heavier reliance on low-level processing of global figures. This study demonstrates that enhanced visual functioning is specific to a cognitive phenotypic subgroup of autistics with superior visuospatial abilities and reinforces the need to address autistic heterogeneity by good cognitive characterization of samples in future studies.

Clinical subtyping using community detection: Limited utility?

OBJECTIVES

To discover psychiatric subtypes, researchers are adopting a method called community detection. This method was not subjected to the same scrutiny in the psychiatric literature as traditional clustering methods. Furthermore, many community detection algorithms have been developed without psychiatric sample sizes and variable numbers in mind. We aim to provide clarity to researchers on the utility of this method.

METHODS

We provide an introduction to community detection algorithms, specifically describing the crucial differences between correlation-based and distance-based community detection. We compare community detection results to results of traditional methods in a simulation study representing typical psychiatry settings, using three conceptualizations of how subtypes might differ.

RESULTS

We discovered that the number of recovered subgroups was often incorrect with several community detection algorithms. Correlation-based community detection fared better than distance-based community detection, and performed relatively well with smaller sample sizes. Latent profile analysis was more consistent in recovering subtypes. Whether methods were successful depended on how differences were introduced.

CONCLUSIONS

Traditional methods like latent profile analysis remain reasonable choices. Furthermore, results depend on assumptions and theoretical choices underlying subtyping analyses, which researchers need to consider before drawing conclusions on subtypes. Employing multiple subtyping methods to establish method dependency is recommended.

Functional activity level reported by an informant is an early predictor of Alzheimer's disease

BACKGROUND

Loss of autonomy in day-to-day functioning is one of the feared outcomes of Alzheimer's disease (AD), and relatives may have been worried by subtle behavioral changes in ordinary life situations long before these changes are given medical attention. In the present study, we ask if such subtle changes should be given weight as an early predictor of a future AD diagnosis.

METHODS

Longitudinal data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) were used to define a group of adults with a mild cognitive impairment (MCI) diagnosis remaining stable across several visits (sMCI, n=360; 55-91 years at baseline), and a group of adults who over time converted from having an MCI diagnosis to an AD diagnosis (cAD, n=320; 55-88 years at baseline). Eleven features were used as input in a Random Forest (RF) binary classifier (sMCI vs. cAD) model. This model was tested on an unseen holdout part of the dataset, and further explored by three different permutation-driven importance estimates and a comprehensive post hoc machine learning exploration.

RESULTS

The results consistently showed that measures of daily life functioning, verbal memory function, and a volume measure of hippocampus were the most important predictors of conversion from an MCI to an AD diagnosis. Results from the RF classification model showed a prediction accuracy of around 70% in the test set. Importantly, the post hoc analyses showed that even subtle changes in everyday functioning noticed by a close informant put MCI patients at increased risk for being on a path toward the major cognitive impairment of an AD diagnosis.

CONCLUSION

The results showed that even subtle changes in everyday functioning should be noticed when reported by relatives in a clinical evaluation of patients with MCI. Information of these changes should also be included in future longitudinal studies to investigate different pathways from normal cognitive aging to the cognitive decline characterizing different stages of AD and other neurodegenerative disorders.

Evaluating the predictive performance of subtyping: A criterion for cluster mean-based prediction

Heterogeneity is a frequent issue in population data analyses in medicine, biology, and the social sciences. A common approach for handling heterogeneity is to use a clustering algorithm to group similar samples, considering samples within the same group to be homogeneous. This approach is known as "subtyping" or "subgrouping." Methods for evaluating the validity of subtyping have yet to be fully established. In this study, we propose the cost of cluster mean-based prediction (CCMP) as a metric for evaluating the accuracy of predictions based on subtyping. By selecting the minimum CCMP among several candidate clustering results, the optimal subtype classification in terms of prediction accuracy can be determined. The computational implementation of the CCMP is validated with numerical experiments. We also examine some properties of subtype classification selected by CCMP.

Language in autism: domains, profiles and co-occurring conditions

This article reviews the current knowledge state on pragmatic and structural language abilities in autism and their potential relation to extralinguistic abilities and autistic traits. The focus is on questions regarding autism language profiles with varying degrees of (selective) impairment and with respect to potential comorbidity of autism and language impairment: Is language impairment in autism the co-occurrence of two distinct conditions (comorbidity), a consequence of autism itself (no comorbidity), or one possible combination from a series of neurodevelopmental properties (dimensional approach)? As for language profiles in autism, three main groups are identified, namely, (i) verbal autistic individuals without structural language impairment, (ii) verbal autistic individuals with structural language impairment, and (iii) minimally verbal autistic individuals. However, this tripartite distinction hides enormous linguistic heterogeneity. Regarding the nature of language impairment in autism, there is currently no model of how language difficulties may interact with autism characteristics and with various extralinguistic cognitive abilities. Building such a model requires carefully designed explorations that address specific aspects of language and extralinguistic cognition. This should lead to a fundamental increase in our understanding of language impairment in autism, thereby paving the way for a substantial contribution to the question of how to best characterize neurodevelopmental disorders.

Differentiated Approach to Pharmacotherapy of Autism Spectrum Disorders: Biochemical Aspects

Autism Spectrum Disorders (ASD) are highly heterogeneous neurodevelopmental disorders caused by a complex interaction of numerous genetic and environmental factors and leading to deviations in the nervous system formation at the very early developmental stages. Currently, there are no accepted pharmacological treatments for the so-called core symptoms of ASD, such as social communication disorders and restricted and repetitive behavior patterns. Lack of knowledge about biological basis of ASD, absence of the clinically significant biochemical parameters reflecting abnormalities in the signaling cascades controlling the nervous system development and functioning, and lack of methods for selection of clinically and biologically homogeneous subgroups are considered as causes for the failure of clinical trials of ASD pharmacotherapy. This review considers the possibilities of applying differentiated clinical and biological approaches to the targeted search for ASD pharmacotherapy with emphasis on biochemical markers associated with ASD and attempts to stratify patients by biochemical parameters. The use of such approach as "the target-oriented therapy and assessment of the target status before and during the treatment to identify patients with a positive response to treatment" is discussed using the published results of clinical trials as examples. It is concluded that identification of biochemical parameters for selection of the distinct subgroups among the ASD patients requires research on large samples reflecting clinical and biological diversity of the patients with ASD, and use of unified approaches for such studies. An integrated approach, including clinical observation, clinical-psychological assessment of the patient behavior, study of medical history and description of individual molecular profiles should become a new strategy for stratifying patients with ASD for clinical pharmacotherapeutic trials, as well as for evaluating their efficiency.

Translating precision medicine for autism spectrum disorder: A pressing need

Autism spectrum disorder (ASD) is a heterogenous group of neurodevelopmental disorders (NDDs) with a high unmet medical need. Currently, ASD is diagnosed according to behavior-based criteria that overlook clinical and genomic heterogeneity, thus repeatedly resulting in failed clinical trials. Here, we summarize the scientific evidence pointing to the pressing need to create a precision medicine framework for ASD and other NDDs. We discuss the role of omics and systems biology to characterize more homogeneous disease subtypes with different underlying pathophysiological mechanisms and to determine corresponding tailored treatments. Finally, we provide recent initiatives towards tackling the complexity in NDDs for precision medicine and cost-effective drug discovery.

Transdiagnostic profiles of behaviour and communication relate to academic and socioemotional functioning and neural white matter organisation

BACKGROUND

Behavioural and language difficulties co-occur in multiple neurodevelopmental conditions. Our understanding of these problems has arguably been slowed by an overreliance on study designs that compare diagnostic groups and fail to capture the overlap across different neurodevelopmental disorders and the heterogeneity within them.

METHODS

We recruited a large transdiagnostic cohort of children with complex needs (N = 805) to identify distinct subgroups of children with common profiles of behavioural and language strengths and difficulties. We then investigated whether and how these data-driven groupings could be distinguished from a comparison sample (N = 158) on measures of academic and socioemotional functioning and patterns of global and local white matter connectome organisation. Academic skills were assessed via standardised measures of reading and maths. Socioemotional functioning was captured by the parent-rated version of the Strengths and Difficulties Questionnaire.

RESULTS

We identified three distinct subgroups of children, each with different levels of difficulties in structural language, pragmatic communication, and hot and cool executive functions. All three subgroups struggled with academic and socioemotional skills relative to the comparison sample, potentially representing three alternative but related developmental pathways to difficulties in these areas. The children with the weakest language skills had the most widespread difficulties with learning, whereas those with more pronounced difficulties with hot executive skills experienced the most severe difficulties in the socioemotional domain. Each data-driven subgroup could be distinguished from the comparison sample based on both shared and subgroup-unique patterns of neural white matter organisation. Children with the most pronounced deficits in language, cool executive, or hot executive function were differentiated from the comparison sample by altered connectivity in predominantly thalamocortical, temporal-parietal-occipital, and frontostriatal circuits, respectively.

CONCLUSIONS

These findings advance our understanding of commonly co-morbid behavioural and language problems and their relationship to behavioural outcomes and neurobiological substrates.

Leveraging Machine Learning for Gaining Neurobiological and Nosological Insights in Psychiatric Research

Much attention is currently devoted to developing diagnostic classifiers for mental disorders. Complementing these efforts, we highlight the potential of machine learning to gain biological insights into the psychopathology and nosology of mental disorders. Studies to this end have mainly used brain imaging data, which can be obtained noninvasively from large cohorts and have repeatedly been argued to reveal potentially intermediate phenotypes. This may become particularly relevant in light of recent efforts to identify magnetic resonance imaging-derived biomarkers that yield insight into pathophysiological processes as well as to refine the taxonomy of mental illness. In particular, the accuracy of machine learning models may be used as dependent variables to identify features relevant to pathophysiology. Moreover, such approaches may help disentangle the dimensional (within diagnosis) and often overlapping (across diagnoses) symptomatology of psychiatric illness. We also point out a multiview perspective that combines data from different sources, bridging molecular and system-level information. Finally, we summarize recent efforts toward a data-driven definition of subtypes or disease entities through unsupervised and semisupervised approaches. The latter, blending unsupervised and supervised concepts, may represent a particularly promising avenue toward dissecting heterogeneous categories. Finally, we raise several technical and conceptual aspects related to the reviewed approaches. In particular, we discuss common pitfalls pertaining to flawed input data or analytic procedures that would likely lead to unreliable outputs.

Modern views of machine learning for precision psychiatry

In light of the National Institute of Mental Health (NIMH)'s Research Domain Criteria (RDoC), the advent of functional neuroimaging, novel technologies and methods provide new opportunities to develop precise and personalized prognosis and diagnosis of mental disorders. Machine learning (ML) and artificial intelligence (AI) technologies are playing an increasingly critical role in the new era of precision psychiatry. Combining ML/AI with neuromodulation technologies can potentially provide explainable solutions in clinical practice and effective therapeutic treatment. Advanced wearable and mobile technologies also call for the new role of ML/AI for digital phenotyping in mobile mental health. In this review, we provide a comprehensive review of ML methodologies and applications by combining neuroimaging, neuromodulation, and advanced mobile technologies in psychiatry practice. We further review the role of ML in molecular phenotyping and cross-species biomarker identification in precision psychiatry. We also discuss explainable AI (XAI) and neuromodulation in a closed human-in-the-loop manner and highlight the ML potential in multi-media information extraction and multi-modal data fusion. Finally, we discuss conceptual and practical challenges in precision psychiatry and highlight ML opportunities in future research.

Functional Connectome-Based Predictive Modeling in Autism

Autism is a heterogeneous neurodevelopmental condition, and functional magnetic resonance imaging-based studies have helped advance our understanding of its effects on brain network activity. We review how predictive modeling, using measures of functional connectivity and symptoms, has helped reveal key insights into this condition. We discuss how different prediction frameworks can further our understanding of the brain-based features that underlie complex autism symptomatology and consider how predictive models may be used in clinical settings. Throughout, we highlight aspects of study interpretation, such as data decay and sampling biases, that require consideration within the context of this condition. We close by suggesting exciting future directions for predictive modeling in autism.

Differential neural correlates underlying mental rotation processes in two distinct cognitive profiles in autism

Enhanced visuospatial abilities characterize the cognitive profile of a subgroup of autistics. However, the neural correlates underlying such cognitive strengths are largely unknown. Using functional magnetic resonance imaging (fMRI), we investigated the neural underpinnings of superior visuospatial functioning in different autistic subgroups. Twenty-seven autistic adults, including 13 with a Wechsler's Block Design peak (AUTp) and 14 without (AUTnp), and 23 typically developed adults (TYP) performed a classic mental rotation task. As expected, AUTp participants were faster at the task compared to TYP. At the neural level, AUTp participants showed enhanced bilateral parietal and occipital activation, stronger occipito-parietal and fronto-occipital connectivity, and diminished fronto-parietal connectivity compared to TYP. On the other hand, AUTnp participants presented greater activation in right and anterior regions compared to AUTp. In addition, reduced connectivity between occipital and parietal regions was observed in AUTnp compared to AUTp and TYP participants. A greater reliance on posterior regions is typically reported in the autism literature. Our results suggest that this commonly reported finding may be specific to a subgroup of autistic individuals with enhanced visuospatial functioning. Moreover, this study demonstrated that increased occipito-frontal synchronization was associated with superior visuospatial abilities in autism. This finding contradicts the long-range under-connectivity hypothesis in autism. Finally, given the relationship between distinct cognitive profiles in autism and our observed differences in brain functioning, future studies should provide an adequate characterization of the autistic subgroups in their research. The main limitations are small sample sizes and the inclusion of male-only participants.

Re-Examining Labels in Neurocognitive Research: Evidence from Bilingualism and Autism as Spectrum-Trait Cases

Despite the fact that the urge to investigate bilingualism and neurodevelopmental disorders as continuous indices rather than categorical ones has been well-voiced among researchers with respect to research methodological approaches, in the recent literature, when it comes to examining language, cognitive skills and neurodivergent characteristics, it is still the case that the most prevalent view is the categorisation of adults or children into groups. In other words, there is a categorisation of individuals, e.g., monolingual vs. bilingual children or children with typical and atypical/non-typical/non-neurotypical development. We believe that this labelling is responsible for the conflicting results that we often come across in studies. The aim of this review is to bring to the surface the importance of individual differences through the study of relevant articles conducted in bilingual children and children with autism, who are ideal for this study. We concur with researchers who already do so, and we further suggest moving away from labels and instead shift towards the view that not everything is either white or black. We provide suggestions as to how this shift could be implemented in research, while mostly aiming at starting a discourse rather than offering a definite path.

Altered brain network topology in children with auditory processing disorder: A resting-state multi-echo fMRI study

Children with auditory processing disorder (APD) experience hearing difficulties, particularly in the presence of competing sounds, despite having normal audiograms. There is considerable debate on whether APD symptoms originate from bottom-up (e.g., auditory sensory processing) and/or top-down processing (e.g., cognitive, language, memory). A related issue is that little is known about whether functional brain network topology is altered in APD. Therefore, we used resting-state functional magnetic resonance imaging data to investigate the functional brain network organization of 57 children from 8 to 14 years old, diagnosed with APD (n = 28) and without hearing difficulties (healthy control, HC; n = 29). We applied complex network analysis using graph theory to assess the whole-brain integration and segregation of functional networks and brain hub architecture. Our results showed children with APD and HC have similar global network properties -i.e., an average of all brain regions- and modular organization. Still, the APD group showed different hub architecture in default mode-ventral attention, somatomotor and frontoparietal-dorsal attention modules. At the nodal level -i.e., single-brain regions-, we observed decreased participation coefficient (PC - a measure quantifying the diversity of between-network connectivity) in auditory cortical regions in APD, including bilateral superior temporal gyrus and left middle temporal gyrus. Beyond auditory regions, PC was also decreased in APD in bilateral posterior temporo-occipital cortices, left intraparietal sulcus, and right posterior insular cortex. Correlation analysis suggested a positive association between PC in the left parahippocampal gyrus and the listening-in-spatialized-noise -sentences task where APD children were engaged in auditory perception. In conclusion, our findings provide evidence of altered brain network organization in children with APD, specific to auditory networks, and shed new light on the neural systems underlying children's listening difficulties.

Predicting the future of neuroimaging predictive models in mental health

Predictive modeling using neuroimaging data has the potential to improve our understanding of the neurobiology underlying psychiatric disorders and putatively information interventions. Accordingly, there is a plethora of literature reviewing published studies, the mathematics underlying machine learning, and the best practices for using these approaches. As our knowledge of mental health and machine learning continue to evolve, we instead aim to look forward and "predict" topics that we believe will be important in current and future studies. Some of the most discussed topics in machine learning, such as bias and fairness, the handling of dirty data, and interpretable models, may be less familiar to the broader community using neuroimaging-based predictive modeling in psychiatry. In a similar vein, transdiagnostic research and targeting brain-based features for psychiatric intervention are modern topics in psychiatry that predictive models are well-suited to tackle. In this work, we target an audience who is a researcher familiar with the fundamental procedures of machine learning and who wishes to increase their knowledge of ongoing topics in the field. We aim to accelerate the utility and applications of neuroimaging-based predictive models for psychiatric research by highlighting and considering these topics. Furthermore, though not a focus, these ideas generalize to neuroimaging-based predictive modeling in other clinical neurosciences and predictive modeling with different data types (e.g., digital health data).

Identifying neuroanatomical and behavioral features for autism spectrum disorder diagnosis in children using machine learning

Autism spectrum disorder (ASD) is a neurodevelopmental disorder that can cause significant social, communication, and behavioral challenges. Diagnosis of ASD is complicated and there is an urgent need to identify ASD-associated biomarkers and features to help automate diagnostics and develop predictive ASD models. The present study adopts a novel evolutionary algorithm, the conjunctive clause evolutionary algorithm (CCEA), to select features most significant for distinguishing individuals with and without ASD, and is able to accommodate datasets having a small number of samples with a large number of feature measurements. The dataset is unique and comprises both behavioral and neuroimaging measurements from a total of 28 children from 7 to 14 years old. Potential biomarker candidates identified include brain volume, area, cortical thickness, and mean curvature in specific regions around the cingulate cortex, frontal cortex, and temporal-parietal junction, as well as behavioral features associated with theory of mind. A separate machine learning classifier (i.e., k-nearest neighbors algorithm) was used to validate the CCEA feature selection and for ASD prediction. Study findings demonstrate how machine learning tools might help move the needle on improving diagnostic and predictive models of ASD.

Prenatal Exposure to Air Pollution and Early Life Stress Effects on Hippocampal Subregional Volumes and Associations with Visual-Spatial Reasoning

Background

Children from economically distressed families and neighborhoods are at risk for stress and pollution exposure and potential neurotoxic sequelae. We examine dimensions of early-life stress affecting hippocampal volumes, how prenatal exposure to air pollution might magnify these effects, and associations between hippocampal volumes and visuospatial reasoning.

Methods

Fifty-three Hispanic/Latinx and/or Black children of ages 7 to 9 years were recruited from a longitudinal birth cohort for magnetic resonance imaging and cognitive assessment. Exposure to airborne polycyclic aromatic hydrocarbons was measured during the third trimester of pregnancy. Maternal report of psychosocial stress was collected at child age 5 and served as measures of early-life stress. Whole hippocampus and subfield volumes were extracted using FreeSurfer. Wechsler performance IQ measured visuospatial reasoning.

Results

Maternal perceived stress associated with smaller right hippocampal volume among their children (B = −0.57, t₃₄ = −3.05, 95% CI, −0.95 to −0.19). Prenatal polycyclic aromatic hydrocarbon moderated the association between maternal perceived stress and right CA1, CA3, and CA4/dentate gyrus volumes (B ≥ 0.68, t₃₃ ≥ 2.17) such that higher prenatal polycyclic aromatic hydrocarbon exposure magnified negative associations between stress and volume, whereas this was buffered at lower exposure. Right CA3 and CA4/dentate gyrus volumes (B ≥ 0.35, t₃₃ > 2.16) were associated with greater performance IQ.

Conclusions

Prenatal and early-life exposures to chemical and social stressors are likely compounding. Socioeconomic deprivation and disparities increase risk of these exposures that exert critical neurobiological effects. Developing deeper understandings of these complex interactions will facilitate more focused public health strategies to protect and foster the development of children at greatest risk of mental and physical effects associated with poverty.

Diagnosis-informed connectivity subtyping discovers subgroups of autism with reproducible symptom profiles

Clinical heterogeneity has been one of the main barriers to develop effective biomarkers and therapeutic strategies in autism spectrum disorder (ASD). Recognizing this challenge, much effort has been made in recent neuroimaging studies to find biologically more homogeneous subgroups (called 'neurosubtypes') in autism. However, most approaches have rarely evaluated how much the employed features in subtyping represent the core anomalies of ASD, obscuring its utility in actual clinical diagnosis. To address this, we combined two data-driven methods, 'connectome-based gradient' and 'functional random forest', collectively allowing to discover reproducible neurosubtypes based on resting-state functional connectivity profiles that are specific to ASD. Indeed, the former technique provides the features (as input for subtyping) that effectively summarize whole-brain connectome variations in both normal and ASD conditions, while the latter leverages a supervised random forest algorithm to inform diagnostic labels to clustering, which makes neurosubtyping driven by the features of ASD core anomalies. Applying this framework to the open-sharing Autism Brain Imaging Data Exchange repository data (discovery, n = 103/108 for ASD/typically developing [TD]; replication, n = 44/42 for ASD/TD), we found three dominant subtypes of functional gradients in ASD and three subtypes in TD. The subtypes in ASD revealed distinct connectome profiles in multiple brain areas, which are associated with different Neurosynth-derived cognitive functions previously implicated in autism studies. Moreover, these subtypes showed different symptom severity, which degree co-varies with the extent of functional gradient changes observed across the groups. The subtypes in the discovery and replication datasets showed similar symptom profiles in social interaction and communication domains, confirming a largely reproducible brain-behavior relationship. Finally, the connectome gradients in ASD subtypes present both common and distinct patterns compared to those in TD, reflecting their potential overlap and divergence in terms of developmental mechanisms involved in the manifestation of large-scale functional networks. Our study demonstrated a potential of the diagnosis-informed subtyping approach in developing a clinically useful brain-based classification system for future ASD research.

Reproducible neuroimaging features for diagnosis of autism spectrum disorder with machine learning

Autism spectrum disorder (ASD) is the fourth most common neurodevelopmental disorder, with a prevalence of 1 in 160 children. Accurate diagnosis relies on experts, but such individuals are scarce. This has led to increasing interest in the development of machine learning (ML) models that can integrate neuroimaging features from functional and structural MRI (fMRI and sMRI) to help reveal central nervous system alterations characteristic of ASD. We optimized and compared the performance of 12 of the most popular and powerful ML models. Each was separately trained using 15 different combinations of fMRI and sMRI features and optimized with an unbiased model search. Deep learning models predicted ASD with the highest diagnostic accuracy and generalized well to other MRI datasets. Our model achieves state-of-the-art 80% area under the ROC curve (AUROC) in diagnosis on test data from the IMPAC dataset; and 86% and 79% AUROC on the external ABIDE I and ABIDE II datasets (with further improvement to 93% and 90% after supervised domain adaptation). The highest performing models identified reproducible putative biomarkers for accurate ASD diagnosis in accord with known ASD markers as well as novel cerebellar biomarkers. Such reproducibility lends credence to their tremendous potential for defining and using a set of truly generalizable ASD biomarkers that will advance scientific understanding of neuronal changes in ASD.

Filtering respiratory motion artifact from resting state fMRI data in infant and toddler populations

The importance of motion correction when processing resting state functional magnetic resonance imaging (rs-fMRI) data is well-established in adult cohorts. This includes adjustments based on self-limited, large amplitude subject head motion, as well as factitious rhythmic motion induced by respiration. In adults, such respiration artifact can be effectively removed by applying a notch filter to the motion trace, resulting in higher amounts of data retained after frame censoring (e.g., "scrubbing") and more reliable correlation values. Due to the unique physiological and behavioral characteristics of infants and toddlers, rs-fMRI processing pipelines, including methods to identify and remove colored noise due to subject motion, must be appropriately modified to accurately reflect true neuronal signal. These younger cohorts are characterized by higher respiration rates and lower-amplitude head movements than adults; thus, the presence and significance of comparable respiratory artifact and the subsequent necessity of applying similar techniques remain unknown. Herein, we identify and characterize the consistent presence of respiratory artifact in rs-fMRI data collected during natural sleep in infants and toddlers across two independent cohorts (aged 8-24 months) analyzed using different pipelines. We further demonstrate how removing this artifact using an age-specific notch filter allows for both improved data quality and data retention in measured results. Importantly, this work reveals the critical need to identify and address respiratory-driven head motion in fMRI data acquired in young populations through the use of age-specific motion filters as a mechanism to optimize the accuracy of measured results in this population.

Dissecting the heterogeneous subcortical brain volume of autism spectrum disorder using community detection

Structural brain alterations in autism spectrum disorder (ASD) are heterogeneous, with limited effect sizes overall. In this study, we aimed to identify subgroups in ASD, based on neuroanatomical profiles; we hypothesized that the effect sizes for case/control differences would be increased in the newly defined subgroups. Analyzing a large data set from the ENIGMA-ASD working group (n = 2661), we applied exploratory factor analysis (EFA) to seven subcortical volumes of individuals with and without ASD to uncover the underlying organization of subcortical structures. Based on earlier findings and data availability, we focused on three age groups: boys (<=14 years), male adolescents (15-22 years), and adult men (> = 22 years). The resulting factor scores were used in a community detection (CD) analysis to cluster participants into subgroups. Three factors were found in each subsample; the factor structure in adult men differed from that in boys and male adolescents. From these factors, CD uncovered four distinct communities in boys and three communities in adolescents and adult men, irrespective of ASD diagnosis. The effect sizes for case/control comparisons were more pronounced than in the combined sample, for some communities. A significant group difference in ADOS scores between communities was observed in boys and male adolescents with ASD. We succeeded in stratifying participants into more homogeneous subgroups based on subcortical brain volumes. This stratification enhanced our ability to observe case/control differences in subcortical brain volumes in ASD, and may help to explain the heterogeneity of previous findings in ASD. LAY SUMMARY: Structural brain alterations in ASD are heterogeneous, with overall limited effect sizes. Here we aimed to identify subgroups in ASD based on neuroimaging measures. We tested whether the effect sizes for case/control differences would be increased in the newly defined subgroups. Based on neuroanatomical profiles, we succeeded in stratifying our participants into more homogeneous subgroups. The effect sizes of case/control differences were more pronounced in some subgroups than those in the whole sample.

Application and research progress of machine learning in the diagnosis and treatment of neurodevelopmental disorders in children

The prevalence of neurodevelopment disorders (NDDs) among children has been on the rise. This has affected the health and social life of children. This condition has also imposed a huge economic burden on families and health care systems. Currently, it is difficult to perform early diagnosis of NDDs, which results in delayed intervention. For this reason, patients with NDDs have a prognosis. In recent years, machine learning (ML) technology, which integrates artificial intelligence technology and medicine, has been applied in the early detection and prediction of diseases based on data mining. This paper reviews the progress made in the application of ML in the diagnosis and treatment of NDDs in children based on supervised and unsupervised learning tools. The data reviewed here provide new perspectives on early diagnosis and treatment of NDDs.

Multi-site clustering and nested feature extraction for identifying autism spectrum disorder with resting-state fMRI

Brain functional connectivity (FC) derived from resting-state functional magnetic resonance imaging (rs-fMRI) has been widely employed to study neuropsychiatric disorders such as autism spectrum disorder (ASD). Existing studies usually suffer from (1) significant data heterogeneity caused by different scanners or studied populations in multiple sites, (2) curse of dimensionality caused by millions of voxels in each fMRI scan and a very limited number (tens or hundreds) of training samples, and (3) poor interpretability, which hinders the identification of reproducible disease biomarkers. To this end, we propose a Multi-site Clustering and Nested Feature Extraction (MC-NFE) method for fMRI-based ASD detection. Specifically, we first divide multi-site training data into ASD and healthy control (HC) groups. To model inter-site heterogeneity within each category, we use a similarity-driven multiview linear reconstruction model to learn latent representations and perform subject clustering within each group. We then design a nested singular value decomposition (SVD) method to mitigate inter-site heterogeneity and extract FC features by learning both local cluster-shared features across sites within each category and global category-shared features across ASD and HC groups, followed by a linear support vector machine (SVM) for ASD detection. Experimental results on 609 subjects with rs-fMRI from the ABIDE database with 21 imaging sites suggest that the proposed MC-NFE outperforms several state-of-the-art methods in ASD detection. The most discriminative FCs identified by the MC-NFE are mainly located in default mode network, salience network, and cerebellum region, which could be used as potential biomarkers for fMRI-based ASD analysis.

Cognitive correlates of autism spectrum disorder symptoms

Due to the diverse behavioral presentation of autism spectrum disorder (ASD), identifying ASD subtypes using patterns of cognitive abilities has become an important point of research. Some previous studies on cognitive profiles in ASD suggest that the discrepancy between verbal intelligence quotient (VIQ) and performance IQ (PIQ) is associated with ASD symptoms, while others have pointed to VIQ as the critical predictor. Given that VIQ is a component of the VIQ-PIQ discrepancy, it was unclear which was most driving these associations. This study tested whether VIQ, PIQ, or the VIQ-PIQ discrepancy was most associated with ASD symptoms in children and adults with ASD (N = 527). Using data from the Autism Brain Imaging Data Exchange (ABIDE), we tested the independent contribution of each IQ index and their discrepancy to ASD symptom severity using multiple linear regressions predicting ASD symptoms. VIQ was most associated with lower symptom severity as measured by the Autism Diagnostic Observation Schedule (ADOS) total score, and when VIQ was included in models predicting ASD symptoms, associations with PIQ and IQ discrepancy were not significant. An association between VIQ and ASD communication symptoms drove the association with ASD symptom severity. These results suggest that associations between ASD communication symptoms and IQ discrepancy or PIQ reported in prior studies likely resulted from variance shared with VIQ. Subtyping ASD on the basis of VIQ should be a point of future research, as it may allow for the development of more personalized approaches to intervention. LAY SUMMARY: Previous research on links between autism severity and verbal and nonverbal intelligence has produced mixed results. Our study examined whether verbal intelligence, nonverbal intelligence, or the discrepancy between the two was most related to autism symptoms. We found that higher verbal intelligence was most associated with less severe autism communication symptoms. Given the relevance of verbal intelligence in predicting autism symptom severity, subtyping autism on the basis of verbal intelligence could lead to more personalized treatments.

Subtyping Autism Spectrum Disorder via Joint Modeling of Clinical and Connectomic Profiles

BACKGROUND

Autism Spectrum Disorder (ASD) is a highly heterogeneous developmental disorder with diverse clinical manifestations. Neuroimaging studies have explored functional connectivity (FC) of ASD through resting-state functional MRI studies, however the findings have remained inconsistent, thus reflecting the possibility of multiple subtypes. Identification of the relationship between clinical symptoms and FC measures may help clarify the inconsistencies in earlier findings and advance our understanding of ASD subtypes.

METHODS

Canonical correlation analysis was performed on two-hundred and ten ASD subjects from the Autism Brain Imaging Data Exchange to identify significant linear combinations of resting-state connectomic and clinical profiles of ASD. Then, hierarchical clustering defined ASD subtypes based on distinct brain-behavior relationships. Finally, a support vector machine classifier was used to verify that subtypes were comprised of subjects with distinct clinical and connectivity features.

RESULTS

Three ASD subtypes were identified. Subtype 1 exhibited increased intra-network FC, increased IQ scores and restricted and repetitive behaviors. Subtype 2 was characterized by decreased whole-brain FC and more severe ADI-R and SRS symptoms. Subtype 3 demonstrated mixed FC, low IQ scores, as well as social motivation and verbal deficits. To verify subtype assignment, a multi-class support vector machine using connectomic and clinical profiles yielded an average accuracy of 71.3% and 65.2% respectively for subtype classification, which is significantly higher than chance (33.3%).

CONCLUSION

The present study demonstrates that combining connectomic and behavioral measures is a powerful approach for disease subtyping and suggests that there are ASD subtypes with distinct connectomic and clinical profiles.

Co-activation pattern alterations in autism spectrum disorder-A volume-wise hierarchical clustering fMRI study

INTRODUCTION

There has been a growing effort to characterize the time-varying functional connectivity of resting state (RS) fMRI brain networks (RSNs). Although voxel-wise connectivity studies have examined different sliding window lengths, nonsequential volume-wise approaches have been less common.

METHODS

Inspired by earlier co-activation pattern (CAP) studies, we applied hierarchical clustering (HC) to classify the image volumes of the RS-fMRI data on 28 adolescents with autism spectrum disorder (ASD) and their 27 typically developing (TD) controls. We compared the distribution of the ASD and TD groups' volumes in CAPs as well as their voxel-wise means. For simplification purposes, we conducted a group independent component analysis to extract 14 major RSNs. The RSNs' average z-scores enabled us to meaningfully regroup the RSNs and estimate the percentage of voxels within each RSN for which there was a significant group difference. These results were jointly interpreted to find global group-specific patterns.

RESULTS

We found similar brain state proportions in 58 CAPs (clustering interval from 2 to 30). However, in many CAPs, the voxel-wise means differed significantly within a matrix of 14 RSNs. The rest-activated default mode-positive and default mode-negative brain state properties vary considerably in both groups over time. This division was seen clearly when the volumes were partitioned into two CAPs and then further examined along the HC dendrogram of the diversifying brain CAPs. The ASD group network activations followed a more heterogeneous distribution and some networks maintained higher baselines; throughout the brain deactivation state, the ASD participants had reduced deactivation in 12/14 networks. During default mode-negative CAPs, the ASD group showed simultaneous visual network and either dorsal attention or default mode network overactivation.

CONCLUSION

Nonsequential volume gathering into CAPs and the comparison of voxel-wise signal changes provide a complementary perspective to connectivity and an alternative to sliding window analysis.

Validation strategies for subtypes in psychiatry: A systematic review of research on autism spectrum disorder

Heterogeneity within autism spectrum disorder (ASD) is recognized as a challenge to both biological and psychological research, as well as clinical practice. To reduce unexplained heterogeneity, subtyping techniques are often used to establish more homogeneous subtypes based on metrics of similarity and dissimilarity between people. We review the ASD literature to create a systematic overview of the subtyping procedures and subtype validation techniques that are used in this field. We conducted a systematic review of 156 articles (2001-June 2020) that subtyped participants (range N of studies = 17-20,658), of which some or all had an ASD diagnosis. We found a large diversity in (parametric and non-parametric) methods and (biological, psychological, demographic) variables used to establish subtypes. The majority of studies validated their subtype results using variables that were measured concurrently, but were not included in the subtyping procedure. Other investigations into subtypes' validity were rarer. In order to advance clinical research and the theoretical and clinical usefulness of identified subtypes, we propose a structured approach and present the SUbtyping VAlidation Checklist (SUVAC), a checklist for validating subtyping results.

Markov chain Monte Carlo methods for hierarchical clustering of dynamic causal models

In this article, we address technical difficulties that arise when applying Markov chain Monte Carlo (MCMC) to hierarchical models designed to perform clustering in the space of latent parameters of subject‐wise generative models. Specifically, we focus on the case where the subject‐wise generative model is a dynamic causal model (DCM) for functional magnetic resonance imaging (fMRI) and clusters are defined in terms of effective brain connectivity. While an attractive approach for detecting mechanistically interpretable subgroups in heterogeneous populations, inverting such a hierarchical model represents a particularly challenging case, since DCM is often characterized by high posterior correlations between its parameters. In this context, standard MCMC schemes exhibit poor performance and extremely slow convergence. In this article, we investigate the properties of hierarchical clustering which lead to the observed failure of standard MCMC schemes and propose a solution designed to improve convergence but preserve computational complexity. Specifically, we introduce a class of proposal distributions which aims to capture the interdependencies between the parameters of the clustering and subject‐wise generative models and helps to reduce random walk behaviour of the MCMC scheme. Critically, these proposal distributions only introduce a single hyperparameter that needs to be tuned to achieve good performance. For validation, we apply our proposed solution to synthetic and real‐world datasets and also compare it, in terms of computational complexity and performance, to Hamiltonian Monte Carlo (HMC), a state‐of‐the‐art Monte Carlo technique. Our results indicate that, for the specific application domain considered here, our proposed solution shows good convergence performance and superior runtime compared to HMC.

Understanding Vulnerability and Adaptation in Early Brain Development using Network Neuroscience

Early adversity influences brain development and emerging behavioral phenotypes relevant for psychiatric disorders. Understanding the effects of adversity before and after conception on brain development has implications for contextualizing current public health crises and pervasive health inequities. The use of functional magnetic resonance imaging (fMRI) to study the brain at rest has shifted understanding of brain functioning and organization in the earliest periods of life. Here we review applications of this technique to examine effects of early life stress (ELS) on neurodevelopment in infancy, and highlight targets for future research. Building on the foundation of existing work in this area will require tackling significant challenges, including greater inclusion of often marginalized segments of society, and conducting larger, properly powered studies.

Performance of machine learning classification models of autism using resting-state fMRI is contingent on sample heterogeneity

Autism spectrum disorders (ASDs) are heterogeneous neurodevelopmental conditions. In fMRI studies, including most machine learning studies seeking to distinguish ASD from typical developing (TD) samples, cohorts differing in gender and symptom severity composition are often treated statistically as one "ASD group". Using resting-state functional connectivity (FC) data, we implemented random forest to build diagnostic classifiers in 4 ASD samples including a total of 656 participants (N_ASD = 306, N_TD = 350, ages 6-18). Groups were manipulated to titrate heterogeneity of gender and symptom severity and partially overlapped. Each sample differed on inclusionary criteria: (1) all genders, unrestricted severity range; (2) only male participants, unrestricted severity; (3) all genders, higher severity only; (4) only male participants, higher severity. Each set consisted of 200 participants per group (ASD, TD; matched on age and head motion), 160 for training and 40 for validation. FMRI time series from 237 regions of interest (ROIs) were Pearson correlated in a 237×237 FC matrix and classifiers were built using random forest in training samples. Classification accuracies in validation samples were 62.5%, 65%, 70% and 73.75%, respectively for samples 1-4. Connectivity within cingulo-opercular task control (COTC) network, and between COTC ROIs and default mode and dorsal attention network contributed overall most informative features, but features differed across sets. Findings suggest that diagnostic classifiers vary depending on ASD sample composition. Specifically, greater homogeneity of samples regarding gender and symptom severity enhances classifier performance. However, given the true heterogeneity of ASDs, performance metrics alone may not adequately reflect classifier utility.

Plasma metabolites were associated with spatial working memory in major depressive disorder

Major depressive disorder (MDD) is a common disease with both affective and cognitive disorders. Alterations in metabolic systems of MDD patients have been reported, but the underlying mechanisms still remains unclear. We sought to identify abnormal metabolites in MDD by metabolomics and to explore the association between differential metabolites and neurocognitive dysfunction.Plasma samples from 53 MDD patients and 83 sex-, gender-, BMI-matched healthy controls (HCs) were collected. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) system was then used to detect metabolites in those samples. Two different algorithms were applied to identify differential metabolites in 2 groups. Of the 136 participants, 35 MDD patients and 48 HCs had completed spatial working memory test. Spearman rank correlation coefficient was applied to explore the relationship between differential metabolites and working memory in these 2 groups.The top 5 metabolites which were found in sparse partial least squares-discriminant analysis (sPLS-DA) model and random forest (RF) model were the same, and significant difference was found in 3 metabolites between MDD and HCs, namely, gamma-glutamyl leucine, leucine-enkephalin, and valeric acid. In addition, MDD patients had higher scores in spatial working memory (SWM) between errors and total errors than HCs. Valeric acid was positively correlated with working memory in MDD group.Gamma-glutamyl leucine, leucine-enkephalin, and valeric acid were preliminarily proven to be decreased in MDD patients. In addition, MDD patients performed worse in working memory than HCs. Dysfunction in working memory of MDD individuals was associated with valeric acid.

Moving beyond the mean: Subgroups and dimensions of brain activity and cognitive performance across domains

Human neuroimaging during cognitive tasks has provided unique and important insights into the neurobiology of cognition. However, the vast majority of research relies on group aggregate or average statistical maps of activity, which do not fully capture the rich intersubject variability in brain function. In order to fully understand the neurobiology of cognitive processes, it is necessary to explore the range of variability in activation patterns across individuals. To better characterize individual variability, hierarchical clustering was performed separately on six fMRI tasks in 822 participants from the Human Connectome Project. Across all tasks, clusters ranged from a predominantly 'deactivating' pattern towards a more 'activating' pattern of brain activity, with significant differences in out-of-scanner cognitive test scores between clusters. Cluster stability was assessed via a resampling approach; a cluster probability matrix was generated, as the probability of any pair of participants clustering together when both were present in a random subsample. Rather than forming distinct clusters, participants fell along a spectrum or into pseudo-clusters without clear boundaries. A principal components analysis of the cluster probability matrix revealed three components explaining over 90% of the variance in clustering. Plotting participants in this lower-dimensional 'similarity space' revealed manifolds of variations along an S 'snake' shaped spectrum or a folded circle or 'tortilla' shape. The 'snake' shape was present in tasks where individual variability related to activity along covarying networks, while the 'tortilla' shape represented multiple networks which varied independently.

Integration of brain and behavior measures for identification of data-driven groups cutting across children with ASD, ADHD, or OCD

Autism spectrum disorder (ASD), obsessive-compulsive disorder (OCD) and attention-deficit/hyperactivity disorder (ADHD) are clinically and biologically heterogeneous neurodevelopmental disorders (NDDs). The objective of the present study was to integrate brain imaging and behavioral measures to identify new brain-behavior subgroups cutting across these disorders. A subset of the data from the Province of Ontario Neurodevelopmental Disorder (POND) Network was used including participants with different NDDs (aged 6-16 years) that underwent cross-sectional T1-weighted and diffusion-weighted magnetic resonance imaging (MRI) scanning on the same 3T scanner, and behavioral/cognitive assessments. Similarity Network Fusion was applied to integrate cortical thickness, subcortical volume, white matter fractional anisotropy (FA), and behavioral measures in 176 children with ASD, ADHD or OCD with complete data that passed quality control. Normalized mutual information was used to determine top contributing model features. Bootstrapping, out-of-model outcome measures and supervised machine learning were each used to examine stability and evaluate the new groups. Cortical thickness in socio-emotional and attention/executive networks and inattention symptoms comprised the top ten features driving participant similarity and differences between four transdiagnostic groups. Subcortical volumes (pallidum, nucleus accumbens, thalamus) were also different among groups, although white matter FA showed limited differences. Features driving participant similarity remained stable across resampling, and the new groups showed significantly different scores on everyday adaptive functioning. Our findings open the possibility of studying new data-driven groups that represent children with NDDs more similar to each other than others within their own diagnostic group. Future work is needed to build on this early attempt through replication of the current findings in independent samples and testing longitudinally for prognostic value.

A Review and a Framework of Variables for Defining and Characterizing Tinnitus Subphenotypes

Tinnitus patients can present with various characteristics, such as those related to the tinnitus perception, symptom severity, and pattern of comorbidities. It is speculated that this phenotypic heterogeneity is associated with differences in the underlying pathophysiology and personal reaction to the condition. However, there is as yet no established protocol for tinnitus profiling or subtyping, hindering progress in treatment development. This review summarizes data on variables that have been used in studies investigating phenotypic differences in subgroups of tinnitus, including variables used to both define and compare subgroups. A PubMed search led to the identification of 64 eligible articles. In most studies, variables for subgrouping were chosen by the researchers (hypothesis-driven approach). Other approaches included application of unsupervised machine-learning techniques for the definition of subgroups (data-driven), and subgroup definition based on the response to a tinnitus treatment (treatment response). A framework of 94 variable concepts was created to summarize variables used across all studies. Frequency statistics for the use of each variable concept are presented, demonstrating those most and least commonly assessed. This review highlights the high dimensionality of tinnitus heterogeneity. The framework of variables can contribute to the design of future studies, helping to decide on tinnitus assessment and subgrouping.

Common and unique multimodal covarying patterns in autism spectrum disorder subtypes

BACKGROUND

The heterogeneity inherent in autism spectrum disorder (ASD) presents a substantial challenge to diagnosis and precision treatment. Heterogeneity across biological etiologies, genetics, neural systems, neurocognitive attributes and clinical subtypes or phenotypes has been observed across individuals with ASD.

METHODS

In this study, we aim to investigate the heterogeneity in ASD from a multimodal brain imaging perspective. The Autism Diagnostic Observation Schedule (ADOS) was used as a reference to guide functional and structural MRI fusion. DSM-IV-TR diagnosed Asperger's disorder (n = 79), pervasive developmental disorder-not otherwise specified [PDD-NOS] (n = 58) and Autistic disorder (n = 92) from ABIDE II were used as discovery cohort, and ABIDE I (n = 400) was used for replication.

RESULTS

Dorsolateral prefrontal cortex and superior/middle temporal cortex are the primary common functional-structural covarying cortical brain areas shared among Asperger's, PDD-NOS and Autistic subgroups. Key differences among the three subtypes are negative functional features within subcortical brain areas, including negative putamen-parahippocampus fractional amplitude of low-frequency fluctuations (fALFF) unique to the Asperger's subtype; negative fALFF in anterior cingulate cortex unique to PDD-NOS subtype; and negative thalamus-amygdala-caudate fALFF unique to the Autistic subtype. Furthermore, each subtype-specific brain pattern is correlated with different ADOS subdomains, with social interaction as the common subdomain. The identified subtype-specific patterns are only predictive for ASD symptoms manifested in the corresponding subtypes, but not the other subtypes.

CONCLUSIONS

Although ASD has a common neural basis with core deficits linked to social interaction, each ASD subtype is strongly linked to unique brain systems and subdomain symptoms, which may help to better understand the underlying mechanisms of ASD heterogeneity from a multimodal neuroimaging perspective.

LIMITATIONS

This study is male based, which cannot be generalized to the female or the general ASD population.