Multi-instance Conditional Autoencoder a data driven compression model for strongly correlated datasets

Author(s):  
Jalal Al-afandi ◽  
Andras Horvath
2021 ◽  
Author(s):  
Michael Levine ◽  
Huan Chen ◽  
Ericka Wodka ◽  
Brian Caffo ◽  
Joshua Ewen

Background: There is a substantial base of literature studying the relationship between general intelligence and autism spectrum disorder (ASD). However, the study of this relationship has produced varied results, possibly due to the evolution of diagnostic criteria, measurement, and latent constructs within psychology. We aimed to assess the degree to which core ASD symptom severity relates to aspects of general intellectual functioning, and determine whether this relationship is sensitive to different versions of IQ tests. Method: We conducted a factor analysis on Wechsler Intelligence Scale for Children, fifth edition (WISC-V) (N=83) and Wechsler Intelligence Scale for Children, fourth edition (WISC-IV) (N=131) data in children with ASD and examined the relationship between factors and Autism Diagnostic Observational Schedule (ADOS) performance. Then, we compared the data-driven factor analysis with the manualized IQ indices. We also examined the WISC-IV in a typically developing (TD) cohort (N=209). Results: Results showed that the data-driven factor analysis in TDs was tightly correlated with the manual-derived indices, ADOS scores in children with ASD were poorly correlated with full-scale IQ and manualized indices in WISC-V but were more strongly correlated with IQ test results in the WISC-IV group, and there was weaker correlation between data-driven factors and manualized IQ index scores in the ASD group than in the TD group. Conclusions: While the most recent version of the WISC is less influenced by symptoms of ASD, index-level scores are impacted to the point that the structure of the hierarchy itself differs between groups, possibly due to working memory.


1989 ◽  
Vol 54 (1) ◽  
pp. 101-105 ◽  
Author(s):  
J. Bruce Tomblin ◽  
Cynthia M. Shonrock ◽  
James C. Hardy

The extent to which the Minnesota Child Development Inventory (MCDI), could be used to estimate levels of language development in 2-year-old children was examined. Fifty-seven children between 23 and 28 months were given the Sequenced Inventory of Communication Development (SICD), and at the same time a parent completed the MCDI. In addition the mean length of utterance (MLU) was obtained for each child from a spontaneous speech sample. The MCDI Expressive Language scale was found to be a strong predictor of both the SICD Expressive scale and MLU. The MCDI Comprehension-Conceptual scale, presumably a receptive language measure, was moderately correlated with the SICD Receptive scale; however, it was also strongly correlated with the expressive measures. These results demonstrated that the Expressive Language scale of the MCDI was a valid predictor of expressive language for 2-year-old children. The MCDI Comprehension-Conceptual scale appeared to assess both receptive and expressive language, thus complicating its interpretation.


Sign in / Sign up

Export Citation Format

Share Document