en
Doctoral thesis
Open access
English

Deep Phenotyping to Quantify Social Deficits in Autism Spectrum Disorders (ASD)

ContributorsKojovic, Nada
Number of pages166
Imprimatur date2022
Defense date2022-09-12
Abstract

Autism, a neurodevelopmental disorder characterized by an array of difficulties in social communication and interaction and the presence of repetitive behaviors and interests, is also marked by a striking disturbance in the hard-wired, phylogenetically preserved mechanisms of social orientation. This thesis uses an multimodal approach to explore these disturbances and their implications for early diagnosis and intervention.

This thesis begins by presenting introductory notions helpful for understanding key aspects of the field. Chapter 1 opens with the evolution of diagnostic criteria for autism, followed by detailed discussions on the behavioral phenotype, current diagnostic criteria, and developmental trajectories. We argue for early detection to catalyze diagnostic referrals, enabling early and intensive intervention, essential for young children on the spectrum. Chapter 2 reviews basic concepts regarding the typical functioning of the human visual attentional system, with a special emphasis on social attention from a developmental perspective. Additionally, we present eye-tracking evidence highlighting aberrant social visual engagement as a main pathognomonic feature of the disorder and briefly discuss main brain imaging findings relevant to understanding the complexity of autism's manifestations. Chapter 3 discusses the potential of novel technologies to leverage clinical experience and research in autism, aiming to bring clinical practice closer to more fine-grained, objective, and scalable measures to enable earlier intervention.

In this thesis, we advance the current understanding of autism through various approaches presented in our four studies. Studies I & II describe a novel method, the Proximity Index (PI), developed to quantify moment-to-moment divergences in gazing patterns in children with autism compared to typically developing children. Our goal was to create a measure that allows flexible conceptualization of normative gaze behavior without imposing priors. Normative gaze behavior is learned from the distribution of gaze in typically developing children and used as a reference point to compare with gazing patterns of children with ASD. We test the clinical relevance of this new method by linking it to the behavioral characteristics of children with ASD. Study II zooms out from micro-structures of gaze deployment to explore the dynamics of gazing patterns over several years during childhood. In Study III, we examine how the content of social scenes influences processing in children with autism. Leveraging a small but growing sample of brain images acquired during night MRI scanning sessions, we explore the cerebral bases of atypical social visual engagement as measured by our Proximity Index method. Finally, Study IV revisits classical imaging methods, using video recordings of interaction sessions, to develop automated recognition of autism signs through machine learning.

This work contributes several new insights into understanding the pathognomonic features of ASD by combining various techniques. The complexity of autism manifestations necessitates a plurality of perspectives. Hence, densely sampled, "deep" phenotyping strategies are critical for a more complete understanding of the disorder's manifestations and the mechanisms governing them.

eng
Keywords
  • Autism
  • MRI
  • Eyetracking
  • Visual Attention
  • Development
  • Brain
  • Gaze idiosyncrasy
  • Screening
  • Digital Phenotype
  • Automated prediction
Citation (ISO format)
KOJOVIC, Nada. Deep Phenotyping to Quantify Social Deficits in Autism Spectrum Disorders (ASD). 2022. doi: 10.13097/archive-ouverte/unige:177391
Main files (1)
Secondary files (1)
Identifiers
23views
4downloads

Technical informations

Creation05/28/2024 12:42:32 PM
First validation05/29/2024 7:29:50 AM
Update time05/29/2024 7:29:50 AM
Status update05/29/2024 7:29:50 AM
Last indexation05/29/2024 7:30:07 AM
All rights reserved by Archive ouverte UNIGE and the University of GenevaunigeBlack