Behavioral Modeling Using Deep Neural Network Framework for ASD Diagnosis and Prognosis

Wadhera, Tanu; Kakkar, Deepti; Rani, Rajneesh

doi:10.1002/9781119792345.ch11

Cited by 9 publications

(3 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While measures employed here were able to be classified effectively with a parametric model is encouraging for future multimodal analyses on neuropsychological assessments, non-parametric methods could be advantageous for successful classification using time-dependent data. Indeed, deep learning techniques deployed on EEG signal for ASD classification in other studies have achieved high accuracy and represent path toward automated ASD diagnosis ( Wadhera et al, 2021 ), although the interpretability of deep learning models remains limited. Finally, this sample consists of highly verbal, average IQ ASD participants and most measures are parent-reported, which could have impacted the generalizability of the results.…”

Section: Discussionmentioning

confidence: 99%

Linear discriminant analysis of phenotypic data for classifying autism spectrum disorder by diagnosis and sex

Jacokes

Jack²,

Sullivan³

et al. 2022

Front. Neurosci.

View full text Add to dashboard Cite

Autism Spectrum Disorder (ASD) is a developmental condition characterized by social and communication differences. Recent research suggests ASD affects 1-in-44 children in the United States. ASD is diagnosed more commonly in males, though it is unclear whether this diagnostic disparity is a result of a biological predisposition or limitations in diagnostic tools, or both. One hypothesis centers on the ‘female protective effect,’ which is the theory that females are biologically more resistant to the autism phenotype than males. In this examination, phenotypic data were acquired and combined from four leading research institutions and subjected to multivariate linear discriminant analysis. A linear discriminant model was trained on the training set and then deployed on the test set to predict group membership. Multivariate analyses of variance were performed to confirm the significance of the overall analysis, and individual analyses of variance were performed to confirm the significance of each of the resulting linear discriminant axes. Two discriminant dimensions were identified between the groups: a dimension separating groups by the diagnosis of ASD (LD1: 87% of variance explained); and a dimension reflective of a diagnosis-by-sex interaction (LD2: 11% of variance explained). The strongest discriminant coefficients for the first discriminant axis divided the sample in domains with known differences between ASD and comparison groups, such as social difficulties and restricted repetitive behavior. The discriminant coefficients for the second discriminant axis reveal a more nuanced disparity between boys with ASD and girls with ASD, including executive functioning and high-order behavioral domains as the dominant discriminators. These results indicate that phenotypic differences between males and females with and without ASD are identifiable using parent report measures, which could be utilized to provide additional specificity to the diagnosis of ASD in female patients, potentially leading to more targeted clinical strategies and therapeutic interventions. The study helps to isolate a phenotypic basis for future empirical work on the female protective effect using neuroimaging, EEG, and genomic methodologies.

show abstract

Section: Discussionmentioning

confidence: 99%

Linear discriminant analysis of phenotypic data for classifying autism spectrum disorder by diagnosis and sex

Jacokes

Jack²,

Sullivan³

et al. 2022

Front. Neurosci.

View full text Add to dashboard Cite

show abstract

“…Previous studies on the neurobiology underling the etiology and symptom presentation of ASD have been as heterogeneous and diverse as the behavioral phenotypes of ASD ( Minshew and Williams, 2007 ). Recent neuroimaging studies have delineated the ASD brain as a representation of a typical organization of structural and functional brain networks ( Uddin et al, 2013 ; Wadhera, 2021 ), affected not only by ASD core symptomatology but also by age, sex, ethnicity, and cognitive profile ( Dosenbach et al, 2010 ; Tomasi and Volkow, 2012 ). ASD symptom severity has been reported to be associated with symptom trajectory, intensity of school services (e.g., number of services required), treatment response, and comorbidities ( Zachor and Ben Itzchak, 2010 ; Adams et al, 2014 ; Andersen et al, 2017 ; Rosen et al, 2019 ).…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, determination of ASD severity may assist in planning individualized treatment plans, tracking treatment effects or disease progression, and providing insight into the neural substrates underlying ASD phenotypic heterogeneity ( Moradi et al, 2017 ; Liu and Huang, 2020 ; Wadhera and Kakkar, 2021 ). Machine learning-based predictive modeling has recently been utilized to decode symptom severity from neuroimaging data ( Sui et al, 2020 ; Wadhera et al, 2021 ); however, compared to binary classification, severity prediction may be more challenging as it requires the quantitative estimation of specific scores along a continuous behavioral measure, over a wide range, rather than just determining group membership ( Shen et al, 2017 ; Sui et al, 2020 ). Although these models used neuroimaging measures like cortical thickness ( Sato et al, 2013 ; Moradi et al, 2017 ), surface area ( Pua et al, 2019 ), and functional connectivity ( Uddin, 2014 ; Yahata et al, 2016 ; Lake et al, 2019 ; D’Souza et al, 2020 ; Liu and Huang, 2020 ; Pua et al, 2021 ) as features, putative findings have demonstrated a lack of consistency and reproducibility among them.…”

Section: Introductionmentioning

confidence: 99%

Sparse Hierarchical Representation Learning on Functional Brain Networks for Prediction of Autism Severity Levels

et al. 2022

View full text Add to dashboard Cite

Machine learning algorithms have been widely applied in diagnostic tools for autism spectrum disorder (ASD), revealing an altered brain connectivity. However, little is known about whether an magnetic resonance imaging (MRI)-based brain network is related to the severity of ASD symptoms in a large-scale cohort. We propose a graph convolution neural network-based framework that can generate sparse hierarchical graph representations for functional brain connectivity. Instead of assigning initial features for each node, we utilized a feature extractor to derive node features and the extracted representations can be fed to a hierarchical graph self-attention framework to effectively represent the entire graph. By incorporating connectivity embeddings in the feature extractor, we propose adjacency embedding networks to characterize the heterogeneous representations of the brain connectivity. Our proposed model variants outperform the benchmarking model with different configurations of adjacency embedding networks and types of functional connectivity matrices. Using this approach with the best configuration (SHEN atlas for node definition, Tikhonov correlation for connectivity estimation, and identity-adjacency embedding), we were able to predict individual ASD severity levels with a meaningful accuracy: the mean absolute error (MAE) and correlation between predicted and observed ASD severity scores resulted in 0.96, and r = 0.61 (P < 0.0001), respectively. To obtain a better understanding on how to generate better representations, we investigate the relationships between the extracted feature embeddings and the graph theory-based nodal measurements using canonical correlation analysis. Finally, we visualized the model to identify the most contributive functional connections for predicting ASD severity scores.

show abstract

Utilizing Artificial Intelligence to Support Autism Research

Varghese,

Qaraqe,

Al-Thani

2024

Studies in Neuroscience, Psychology and Behavioral Economics

View full text Add to dashboard Cite

Behavioral Modeling Using Deep Neural Network Framework for ASD Diagnosis and Prognosis

Cited by 9 publications

References 30 publications

Linear discriminant analysis of phenotypic data for classifying autism spectrum disorder by diagnosis and sex

Linear discriminant analysis of phenotypic data for classifying autism spectrum disorder by diagnosis and sex

Sparse Hierarchical Representation Learning on Functional Brain Networks for Prediction of Autism Severity Levels

Utilizing Artificial Intelligence to Support Autism Research

Contact Info

Product

Resources

About