Use this URL to cite or link to this record in EThOS:
Title: An empirical and computational investigation of variable outcomes in Autism Spectrum Disorder
Author: Davis, Rachael
ISNI:       0000 0004 6353 0367
Awarding Body: Birkbeck, University of London
Current Institution: Birkbeck (University of London)
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis had two aims. The first was to investigate variability observed in the profiles of young children with autism spectrum disorder (ASD), and our ability to predict this variability based on measures in infancy. The second aim was to identify the underlying mechanisms that generate this variability. I combined analyses of clinical data sets and data from computational models to investigate the influences shaping atypical developmental trajectories in ASD. The first aim was addressed using secondary data analysis from a prospective longitudinal dataset, the British Autism Study of Infant Siblings. Clinical, behavioural, and parental report data were collected on 54 infants at risk of ASD (by virtue of having an older sibling with the disorder) and 50 low-risk controls at 7, 14, 24 and 36 months. Chapter 2 investigates whether variability differed at a group level, evaluating whether heterogeneity was exaggerated in highrisk groups versus low-risk controls. Cognitive variability scores distinguished infants with ASD at 36 months. Intra-subject variability was then assessed. A more uneven cognitive profile at 24 months was predictive of lower cognitive abilities at 36 months in high-risk infants overall. In Chapter 3, behavioural measures at 14 months were identified as predictors of diagnostic outcome at 36 months in high-risk infants. Initial results highlighted the importance of environmental factors and social and communicative performance. The predictive power of the subsequent statistical regression equations was validated against recently available data from Phase 2 of the BASIS study, with 125 at-risk infants, demonstrating 71% specificity and 81% sensitivity in predicting ASD characteristics at 24 months. In the second half of the thesis, potential mechanisms generating variability in ASD behavioural profiles were investigated via computational modelling. Thomas, Knowland and Karmiloff-Smith (2011) developed a computational model targeting the regressive sub-type of autism based on the hypothesis that regression could be caused by Over-Pruning of brain connectivity. In Chapter 4, this model is extended to capture other observed developmental trajectories in ASD. Regressive and non-regressive subgroups were identified, and each was reliably distinguished by a distinct pattern of neurocomputational parameters. Regression and early onset of pruning were indicative of poorer developmental outcomes overall. Non-regressive subgroups, both typical and atypical, were then used to investigate response to remediation via behavioural intervention. The simulation work represents the first application of populationlevel models of atypical development to intervention. Small but reliable intervention effects were identified, following a discrete phase of intervention. However, the results indicated a limited scope to intervene, with the greater success using compensatory rather than normalisation techniques. The overall results are discussed with reference to the need for convergent methods to shed light on the constraints shaping atypical developmental trajectories in ASD.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available