Validating a Biomarker Panel to Improve Autism Diagnostic Processes V1
Research type
Research Study
Full title
A Validation Study to Assess the Efficacy of Multiple Biomarker Tasks in Supporting Diagnosis of Autism in Young People
IRAS ID
332386
Contact name
Sonia Ponzo
Contact email
Sponsor organisation
Healios
Duration of Study in the UK
4 years, 11 months, 31 days
Research summary
This study aims to develop and validate a series of non-invasive digital biomarker tasks (hereinafter referred to as biomarker panel) designed to support clinicians conducting diagnostic assessments for autism. The panel will help clinicians reach diagnostic outcomes by providing additional, ecologically-valid information about the client undergoing the assessment. The digital biomarker panel will consist of five tasks, designed to be ultimately used conjunctively, presented through a custom research app for use on smartphones. Individual biomarkers are briefly outlined below.
The first biomarker is a voice-based task. Voice pattern differences have been observed between autistic and non-autistic individuals in numerous areas such as: intonation and stress patterns; speech rate; affective quality; and volume control (Patel et al., 2020).
The second biomarker is eye-gaze patterns. Autistics have been found to have an increased focus on non-social stimuli and reduced gaze towards facial areas of interest such as the eyes (Kong et al., 2022).
The third biomarker is a movement task. Autistic individuals have previously appeared to produce more ‘jerky’ movements, and with greater acceleration and faster velocity, compared to neurotypical individuals (Edey et al., 2016).
The fourth biomarker is an executive function task. Autistic and non-autistic individuals have previously shown differences in cognitive executive function, as measured by participants' responses to Tower of London and similar tasks (e.g. Unterrainer et al., 2016; Lai et al., 2017; Pooragha, Kafi and Sotodeh, 2013).
The fifth and final biomarker is an anti-saccades task. Previous studies have found differences in saccade movements between autistics and non-autistics (e.g. Johnson et al., 2016; Caldani et al., 2023).
Data will be used to train Machine Learning algorithms aiming to differentiate between those with and without autism (including those with other, non-autism neurodevelopmental conditions). Metrics including accuracy, sensitivity and specificity, will be computed to evaluate the obtained models.
REC name
London - Central Research Ethics Committee
REC reference
23/LO/0710
Date of REC Opinion
10 Oct 2023
REC opinion
Further Information Favourable Opinion