Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network
Introduction Autism spectrum disorder (ASD) is among the most common and pervasive neurodevelopmental disorders. Yet, despite decades of research, the neurobiology of ASD is still poorly understood, as inconsistent findings preclude the identification of robust and interpretable neurobiological ma...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Cambridge University Press
2021-04-01
|
| Series: | European Psychiatry |
| Subjects: | |
| Online Access: | https://www.cambridge.org/core/product/identifier/S0924933821003977/type/journal_article |
| _version_ | 1797616734919720960 |
|---|---|
| author | K. Supekar S. Ryali R. Yuan D. Kumar C. De Los Angeles V. Menon |
| author_facet | K. Supekar S. Ryali R. Yuan D. Kumar C. De Los Angeles V. Menon |
| author_sort | K. Supekar |
| collection | DOAJ |
| description |
Introduction
Autism spectrum disorder (ASD) is among the most common and pervasive neurodevelopmental disorders. Yet, despite decades of research, the neurobiology of ASD is still poorly understood, as inconsistent findings preclude the identification of robust and interpretable neurobiological markers and predictors of clinical symptoms.
Objectives
Identify robust and interpretable dynamic brain markers that distinguish children with ASD from typically-developing (TD) children and predict clinical symptom severity.
Methods
We leverage multiple functional brain imaging cohorts (ABIDE, Stanford; N = 1004) and exciting recent advances in explainable artificial intelligence (xAI), to develop a novel multivariate time series deep neural network model that extracts informative brain dynamics features that accurately distinguish between ASD and TD children, and predict clinical symptom severity.
Results
Our model achieved consistently high classification accuracies in cross-validation analysis of data from the ABIDE cohort. Crucially, despite the differences in symptom profiles, age, and data acquisition protocols, our model also accurately classified data from an independent Stanford cohort without additional training. xAI analyses revealed that brain features associated with the default mode network, and the human voice/face processing and communication systems, most clearly distinguished ASD from TD children in both cohorts. Furthermore, the posterior cingulate cortex emerged as robust predictor of the severity of social and communication deficits in ASD in both cohorts.
Conclusions
Our findings, replicated across two independent cohorts, reveal robust and neurobiologically interpretable brain features that detect ASD and predict core phenotypic features of ASD, and have the potential to transform our understanding of the etiology and treatment of the disorder.
Disclosure
No significant relationships.
|
| first_indexed | 2024-03-11T07:45:32Z |
| format | Article |
| id | doaj.art-a964b9c3911a4b8c8d21298e5542c3f5 |
| institution | Directory Open Access Journal |
| issn | 0924-9338 1778-3585 |
| language | English |
| last_indexed | 2024-03-11T07:45:32Z |
| publishDate | 2021-04-01 |
| publisher | Cambridge University Press |
| record_format | Article |
| series | European Psychiatry |
| spelling | doaj.art-a964b9c3911a4b8c8d21298e5542c3f52023-11-17T05:07:47ZengCambridge University PressEuropean Psychiatry0924-93381778-35852021-04-0164S145S14510.1192/j.eurpsy.2021.397Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural networkK. Supekar0S. Ryali1R. Yuan2D. Kumar3C. De Los Angeles4V. Menon5Psychiatry And Behavioral Sciences, Stanford University, Stanford, United States of AmericaPsychiatry And Behavioral Sciences, Stanford University, Stanford, United States of AmericaPsychiatry And Behavioral Sciences, Stanford University, Stanford, United States of AmericaPsychiatry And Behavioral Sciences, Stanford University, Stanford, United States of AmericaPsychiatry And Behavioral Sciences, Stanford University, Stanford, United States of AmericaPsychiatry And Behavioral Sciences, Stanford University, Stanford, United States of America Introduction Autism spectrum disorder (ASD) is among the most common and pervasive neurodevelopmental disorders. Yet, despite decades of research, the neurobiology of ASD is still poorly understood, as inconsistent findings preclude the identification of robust and interpretable neurobiological markers and predictors of clinical symptoms. Objectives Identify robust and interpretable dynamic brain markers that distinguish children with ASD from typically-developing (TD) children and predict clinical symptom severity. Methods We leverage multiple functional brain imaging cohorts (ABIDE, Stanford; N = 1004) and exciting recent advances in explainable artificial intelligence (xAI), to develop a novel multivariate time series deep neural network model that extracts informative brain dynamics features that accurately distinguish between ASD and TD children, and predict clinical symptom severity. Results Our model achieved consistently high classification accuracies in cross-validation analysis of data from the ABIDE cohort. Crucially, despite the differences in symptom profiles, age, and data acquisition protocols, our model also accurately classified data from an independent Stanford cohort without additional training. xAI analyses revealed that brain features associated with the default mode network, and the human voice/face processing and communication systems, most clearly distinguished ASD from TD children in both cohorts. Furthermore, the posterior cingulate cortex emerged as robust predictor of the severity of social and communication deficits in ASD in both cohorts. Conclusions Our findings, replicated across two independent cohorts, reveal robust and neurobiologically interpretable brain features that detect ASD and predict core phenotypic features of ASD, and have the potential to transform our understanding of the etiology and treatment of the disorder. Disclosure No significant relationships. https://www.cambridge.org/core/product/identifier/S0924933821003977/type/journal_articleautismbiomarkersbrain dynamicsfMRI |
| spellingShingle | K. Supekar S. Ryali R. Yuan D. Kumar C. De Los Angeles V. Menon Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network European Psychiatry autism biomarkers brain dynamics fMRI |
| title | Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network |
| title_full | Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network |
| title_fullStr | Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network |
| title_full_unstemmed | Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network |
| title_short | Identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time-series deep neural network |
| title_sort | identification of robust and interpretable brain signatures of autism and clinical symptom severity using a dynamic time series deep neural network |
| topic | autism biomarkers brain dynamics fMRI |
| url | https://www.cambridge.org/core/product/identifier/S0924933821003977/type/journal_article |
| work_keys_str_mv | AT ksupekar identificationofrobustandinterpretablebrainsignaturesofautismandclinicalsymptomseverityusingadynamictimeseriesdeepneuralnetwork AT sryali identificationofrobustandinterpretablebrainsignaturesofautismandclinicalsymptomseverityusingadynamictimeseriesdeepneuralnetwork AT ryuan identificationofrobustandinterpretablebrainsignaturesofautismandclinicalsymptomseverityusingadynamictimeseriesdeepneuralnetwork AT dkumar identificationofrobustandinterpretablebrainsignaturesofautismandclinicalsymptomseverityusingadynamictimeseriesdeepneuralnetwork AT cdelosangeles identificationofrobustandinterpretablebrainsignaturesofautismandclinicalsymptomseverityusingadynamictimeseriesdeepneuralnetwork AT vmenon identificationofrobustandinterpretablebrainsignaturesofautismandclinicalsymptomseverityusingadynamictimeseriesdeepneuralnetwork |