Neurodevelopmental heterogeneity and computational approaches for understanding autism

Jacob, Suma; Wolff, Jason J.; Steinbach, Michael; Doyle, Colleen B.; Kumar, Vipan; Elison, Jed T.

doi:10.1038/s41398-019-0390-0

Cited by 71 publications

(52 citation statements)

References 81 publications

(65 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Jacob et al [34] described how the longitudinal model and computational analytics process might boost inferential capacity to identify debilitation early that might or might not exceed formal diagnosis thresholds of ASD. The study focuses on supervised and unsupervised algorithms for detecting the neuro-developmental heterogeneity of ASD.…”

Section: Classifying Asd By Enhanced New Machine Learning Techniquesmentioning

confidence: 99%

A Review of Machine Learning Methods of Feature Selection and Classification for Autism Spectrum Disorder

et al. 2020

View full text Add to dashboard Cite

Autism Spectrum Disorder (ASD), according to DSM-5 in the American Psychiatric Association, is a neurodevelopmental disorder that includes deficits of social communication and social interaction with the presence of restricted and repetitive behaviors. Children with ASD have difficulties in joint attention and social reciprocity, using non-verbal and verbal behavior for communication. Due to these deficits, children with autism are often socially isolated. Researchers have emphasized the importance of early identification and early intervention to improve the level of functioning in language, communication, and well-being of children with autism. However, due to limited local assessment tools to diagnose these children, limited speech-language therapy services in rural areas, etc., these children do not get the rehabilitation they need until they get into compulsory schooling at the age of seven years old. Hence, efficient approaches towards early identification and intervention through speedy diagnostic procedures for ASD are required. In recent years, advanced technologies like machine learning have been used to analyze and investigate ASD to improve diagnostic accuracy, time, and quality without complexity. These machine learning methods include artificial neural networks, support vector machines, a priori algorithms, and decision trees, most of which have been applied to datasets connected with autism to construct predictive models. Meanwhile, the selection of features remains an essential task before developing a predictive model for ASD classification. This review mainly investigates and analyzes up-to-date studies on machine learning methods for feature selection and classification of ASD. We recommend methods to enhance machine learning’s speedy execution for processing complex data for conceptualization and implementation in ASD diagnostic research. This study can significantly benefit future research in autism using a machine learning approach for feature selection, classification, and processing imbalanced data.

show abstract

Section: Classifying Asd By Enhanced New Machine Learning Techniquesmentioning

confidence: 99%

A Review of Machine Learning Methods of Feature Selection and Classification for Autism Spectrum Disorder

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Autism spectrum disorders (ASDs) are a group of neural developmental disorders characterized by repetitive stereotype behaviors and social defects (Takumi et al, 2019). Hundreds of related genetic mutations have been identified in human patients (Jacob et al, 2019). Among these candidate genes, Src-homology domain 3 (SH3) and multiple ankyrin repeat domains 3b ( Shank3b ) is one of the few genes which can cause the core syndrome of ASD at single mutation (Peca et al, 2011; Varghese et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Decrease of GSK-3β Activity in the Anterior Cingulate Cortex of Shank3b−/− Mice Contributes to Synaptic and Social Deficiency

Wang

Hou

et al. 2019

Front. Cell. Neurosci.

View full text Add to dashboard Cite

Social deficiency is one of the core syndromes of autism spectrum disorders (ASD), for which the underlying developmental mechanism still remains elusive. Anterior cingulate cortex (ACC) plays a key role in integrating social information and regulating social behavior. Recent studies have indicated that synaptic dysfunction in ACC is essential for ASD social defects. In the present study, we investigated the development of synapses and the roles of glycogen synthase kinase 3β (GSK-3β), which mediates multiple synaptic signaling pathways in ACC by using Shank3b−/− mice (a widely used ASD mouse model). Our data revealed that Shank3b mutation abolished the social induced c-Fos expression in ACC. From 4 weeks post-birth, neurons in Shank3b−/− ACC exhibited an obvious decrease in spine density and stubby spines. The length and thickness of post-synaptic density (PSD), the expression of vesicular glutamate transporter 2 (vGlut2) and glutamate receptor 2 (GluR2), and the frequency of miniature excitatory post-synaptic currents (mEPSCs) were significantly reduced in Shank3b−/−ACC. Interestingly, the levels of phosphorylated GSK-3β (Ser9), which inhibits the activity of GSK-3β, decreased along the same time course as the levels of GluR2 increased in ACC during development. Shank3b mutation leads to a dramatic increase of pGSK-3β (Ser9), and decrease of pPSD95 (a substrate of GSK-3β) and GluR2. Local delivery of AAV expressing constitutively active GSK-3β restored the expression of GluR2, increased the spine density and the number of mature spines. More importantly, active GSK-3β significantly promoted the social activity of Shank3b−/− mice. These data, in together, indicate that decrease of GSK-3β activity in ACC may contribute to the synaptic and social defects of Shank3b−/− mice. Enhancing GSK-3β activity may be utilized to treat ASD in the future.

show abstract

“…ASD is characterized by a heterogeneity of symptoms and neurobiological endophenotypes ( Nunes et al, 2019 ; Dickie et al, 2018 ; Jacob et al, 2019 ) among the affected individuals. Data driven, unsupervised subtyping appears as a natural approach to decompose the heterogeneity in ASD and to identify subtypes of functional brain connectivity.…”

Section: Discussionmentioning

confidence: 99%

“…Autism spectrum disorder (ASD) is a prevalent neurodevelopmental condition of impaired social communication and restrictive behaviour, diagnosed in about 1% of children ( Lai et al, 2014 ; Baio et al, 2018 ), that is associated with extensive heterogeneity of behavioural symptoms and neuro-biological endophenotypes ( Jacob et al, 2019 ; Lombardo et al, 2019 ). Functional magnetic resonance imaging (fMRI) has emerged as a promising technology to identify potential biomarkers of functional connectivity (FC) in ASD and other psychiatric disorders ( Castellanos et al, 2013 ).…”

Section: Introductionmentioning

confidence: 99%

Subtypes of functional connectivity associate robustly with ASD diagnosis

Orban

et al. 2020

Preprint

View full text Add to dashboard Cite

Our understanding of the changes in functional brain organization in autism is 17 hampered by the extensive heterogeneity that characterizes this neurodevelopmental disorder. 18 Data driven clustering offers a straightforward way to decompose this heterogeneity into 19 subtypes of distinguishable connectivity types and promises an unbiased framework to 20 investigate behavioural symptoms and causative genetic factors. Yet the robustness and 21 generalizability of these imaging subtypes is unknown. Here, we show that unsupervised 22 functional connectivity subtypes are moderately associated with the clinical diagnosis of autism, 23 and that these associations generalize to independent replication data. We found that subtypes 24 identified robust patterns of functional connectivity, but that a discrete assignment of individuals 25 to these subtypes was not supported by the data. Our results support the use of data driven 26 subtyping as a data dimensionality reduction technique, rather than to establish clinical 27 categories. 28 29 Introduction 30 Autism spectrum disorder (ASD) is a prevalent neurodevelopmental condition of impaired social 31 communication and restrictive behaviour, diagnosed in about 1% of children (Lai et al., 2014; Baio 32 et al. , 2018), that is associated with extensive heterogeneity of behavioural symptoms and neuro-33 biological endophenotypes (Jacob et al., 2019; Lombardo et al., 2019). Functional magnetic reso-34 nance imaging (fMRI) has emerged as a promising technology to identify potential biomarkers of 35 functional connectivity (FC) in ASD and other psychiatric disorders (Castellanos et al., 2013). ever, efforts to characterize the functional brain organization in ASD have so far largely focused 37 on case-control comparisons, thus ignoring the presumed heterogeneity of FC alterations (Nunes 38 et al., 2019; Hahamy et al., 2015). 39 Data driven cluster analysis has long been proposed as a solution to decompose the hetero-40 1 of 23 Manuscript submitted to eLife geneity of behavioural symptoms in ASD into distinct subtypes (Eaves et al., 1994; Beglinger and 41 Smith, 2001), but these subtypes have proven difficult to distinguish in clinical practice (Lord et al., 42 2012) and were recently abandoned in favour of the broader concept of an autism spectrum (Amer-43 ican Psychiatric Association. and DSM-5 Task Force., 2013). The lack of progress towards repro-44 ducible, brain based biomarkers of ASD (Lombardo et al., 2019) has renewed interest in clustering 45 methods to decompose the heterogeneity of brain alterations into distinct subtypes that are hy-46 pothesized to underlie the multitude of behavioural symptoms. 48 neurobiological heterogeneity in ASD and relate it to behavioural symptoms (Hong et al., 2019). 49 Early work on subcortical volume alterations in ASD distinguished four subtypes, but did not find 50 significant differences of behavioural symptoms between them (Hrdlicka et al., 2005). A more re-51 cent multi-modal analysis distinguished th...

show abstract

Neurodevelopmental heterogeneity and computational approaches for understanding autism

Cited by 71 publications

References 81 publications

A Review of Machine Learning Methods of Feature Selection and Classification for Autism Spectrum Disorder

A Review of Machine Learning Methods of Feature Selection and Classification for Autism Spectrum Disorder

Decrease of GSK-3β Activity in the Anterior Cingulate Cortex of Shank3b−/− Mice Contributes to Synaptic and Social Deficiency

Subtypes of functional connectivity associate robustly with ASD diagnosis

Contact Info

Product

Resources

About