Related Experiment Video
Updated: Oct 17, 2025
12:21
Probing the Brain in Autism Using fMRI and Diffusion Tensor Imaging
Published on: September 12, 2011
25.4K
Imputing cognitive impairment in SPARK, a large autism cohort
Chang Shu1,2, LeeAnne Green Snyder3, Yufeng Shen2,4
1Department of Pediatrics, Columbia University Irving Medical Center, New York, New York, USA.
Summary
Machine learning accurately predicts cognitive impairment in children with autism using parent-reported data when standard testing is unavailable. This method aids large-scale autism research by estimating cognitive ability for better treatment development.
Area of Science:
- Neuroscience
- Genetics
- Developmental Psychology
Background:
- Dissecting autism subtypes requires diverse cohorts, with intellectual disability as a key endophenotype.
- Current cognitive assessments are often infeasible for large-scale autism studies.
- Accurate cognitive profiling is crucial for understanding autism heterogeneity and tailoring interventions.
Purpose of the Study:
- To develop and evaluate machine learning models for predicting cognitive impairment in children with autism using parent-reported data.
- To assess the feasibility of imputing cognitive ability in large autism cohorts where standardized testing is not available.
- To identify key predictive features for cognitive impairment in autism.
Main Methods:
- Five machine learning models were developed to predict cognitive impairment (FSIQ<80, FSIQ<70) and FSIQ scores.
- Models utilized parent-reported online survey data from 521 children with autism in the SPARK cohort.
- Model performance was evaluated in an independent set (n=1346) with up to 70% missing data, comparing predictions against clinical IQ data.
Main Results:
- The elastic-net model demonstrated strong performance in imputing cognitive impairment (FSIQ<80) with an AUC of 0.876, sensitivity of 0.772, and specificity of 0.803.
- Key predictive features included parent-reported language and cognitive levels, age at autism diagnosis, and history of services.
- Accurate predictions were also achieved for FSIQ<70 and FSIQ scores, indicating the model's generalizability.
Conclusions:
- Cognitive levels in children with autism can be accurately imputed using commonly collected parent-reported data and machine learning.
- This approach provides a viable method for estimating cognitive ability in large-scale autism studies where psychometric testing is impractical.
- The developed model supports research aimed at understanding autism heterogeneity and developing targeted treatments for individuals with varying cognitive profiles.
More Related Videos
Related Concept Videos
01:19Autism Spectrum Disorder
485
Autism spectrum disorder (ASD) is a neurodevelopmental condition marked by persistent deficits in social communication and interaction alongside restrictive and repetitive behaviors or interests. ASD is sometimes accompanied by intellectual impairment.
These core symptoms manifest differently among individuals, ranging from mild to severe. The disorder's complexity extends beyond its clinical presentation, encompassing a diverse range of biological, cognitive, and sociocultural influences.
These core symptoms manifest differently among individuals, ranging from mild to severe. The disorder's complexity extends beyond its clinical presentation, encompassing a diverse range of biological, cognitive, and sociocultural influences.
485
01:29Intellectual Disability
215
Intellectual disability (ID) is a neurodevelopmental condition characterized by deficits in intellectual and adaptive functioning that manifest during the developmental period. This condition encompasses challenges in reasoning, memory, problem-solving, and learning, accompanied by impairments in everyday life skills, such as communication, self-care, and social interactions. Intellectual disability affects approximately 1% of the population in the United States, impacting an estimated 5...
215