Early Detection of Autism Using a Mobile Application: Asdetect

Background: Early detection of autism is critical as it provides access to early intervention, improving children’s developmental outcomes and quality of life, and decreasing family stress. Over the past 10 years, two large-scale community-based studies for the early detection of autism have been conducted within the Victorian Maternal and Child Health (MCH) system in Australia, using Social Attention and Communication Surveillance; SACS; Barbaro & Dissanayake, 2010; 2013). Monitoring over 30,000 children, the SACS is the most accurate and sensitive early detection method for autism currently available, with 81% positive predictive value (PPV), and estimated sensitivity and specificity of 83.8% and 99.8%. To make this research accessible to the world’s population, including Lower Middle Income Countries (LMICs), an app based on the SACS was developed called ASDetect.

ASDetect (asdetect.org) is a free mobile application for the early detection of autism (11-30 months) that incorporates a modified version of the SACS. Short videos demonstrate key social-communication behaviours followed by a question, with automatic calculation of a child’s ‘likelihood’ for autism (high/low).

Objectives: To determine the psychometric properties of ASDetect in sensitively and accurately identifying children with autism, and to evaluate its acceptability amongst parent users and professionals.

Methods: Parents of children aged between 11-30 months are invited to participate via their MCH nurse, or via social media/word of mouth. Parents register their child’s details on a webpage, download ASDetect, and complete an assessment. All children at ‘high likelihood’ for autism, and a small percentage of children at ‘low likelihood’ for autism, are invited for a free diagnostic assessment by the ASDetect team. Children are assessed at intake and every 6 months until 24-months, with the ADOS-Toddler, Mullen Scales of Early Learning, and ADI-R administered.

Results: To date, 175 parents have registered, with 127 completing an assessment. 26 children (20.5%) were identified at ‘high likelihood’ for autism, with 12 of the 13 children assessed thus far meeting criteria for autism; the remaining child had a developmental/language delay. Updated results to May 2018 will be presented at the conference.

Conclusions: These preliminary data show very promising results for ASDetect’s positive predictive value (PPV; accuracy) in identifying children with autism. However, more data are needed to determine its overall psychometric properties, and acceptability by parents and professionals, which will continue to be collected until 2019, with all children followed up to 30-months-of-age.