Photo credit: Duke University

An autism spectrum disorder can be diagnosed as early as six to twelve months of age. The American Academy of Pediatrics recommends examining all children between the ages of twelve and eighteen months.

Most diagnoses, however, are made after the age of 4, and later detection makes treatment more difficult and expensive.

One in 40 children will be diagnosed with Autism Spectrum Disorder, and Duke currently cares for about 3,000 ASD patients per year. To improve care for patients with ASD, Duke researchers have worked to develop a data science approach to early detection.

Geraldine Dawson, William Cleland Distinguished Professor in the Department of Psychiatry and Behavioral Sciences and Director of the Duke Center for Autism and Brain Development, and Dr. Matthew Engelhard, Conners Fellow on Digital Health in Psychiatry and Behavioral Sciences, recently noted on advances are being made to improve ASD detection and better understand symptoms.

The earlier ASD is detected, the easier and cheaper the treatment is. Children with ASD face challenges in their learning and social environment.

However, ASD varies greatly from case to case. For most people, ASD makes it difficult to find their way around the social world, and those diagnosed often have difficulty understanding facial expressions, maintaining eye contact, and developing strong relationships with their peers.

Leveraging data science for the early detection of autism

Photo credit: Duke University

However, ASD also has many positive qualities, and autistic children often display unique skills and talents. Obtaining a diagnosis is important for people with ASD so that they can receive study accommodation and ensure that those around them encourage growth.

Because early detection is so helpful, researchers began to ask: Can digital behavioral assessments improve our ability to look for neurodevelopmental disorders and monitor patient outcomes?

The current approach to detecting ASD is through questionnaires given to parents. However, there are many problems with this detection method, such as: B. literacy and language barriers, as well as the requirement that caregivers have knowledge of child development. Recent studies have shown that digital assessments can potentially address these challenges by providing direct observation of the child’s behavior as well as the ability to grasp the dynamics of behavior and gather more data about autism.

“Our goal is to reduce the differences in access to screening and enable earlier detection of ASD by developing digital behavioral screening tools that are scalable, doable and more accurate than current pencil and paper questionnaires that are standard of care, “says Dr. Geraldine Dawson.

Leveraging data science for the early detection of autism

Photo credit: Duke University

Guillermo Sapiro, a distinguished professor of electrical engineering and information technology to James B. Duke, and his team have developed an app to do just that.

In the app, videos are shown to the child on an iPad or iPhone, which trigger the child’s reaction to various stimuli. These are the same games and stimuli typically used in diagnostic ASD assessments in the clinic. While they are watching and interacting, the child’s behavior is measured with the selfie camera on the iPhone / iPad. Some behavioral symptoms can be noticed as early as six months of age, such as: B .: Not so much attention to people, decreased affective expression, early motor differences and a lack of orientation on the name.

In the proof-of-concept study, computers were programmed to recognize a child’s reaction to hearing their name. The child’s name was mentioned three times by the examiner while films were being shown. Infants with ASD showed about a second of latency in their responses.

Another study used eye monitoring on an iPhone. Almost a thousand toddlers were presented with a split screen with a person on one side of the screen and toys on the other. Typical toddlers alternated their gaze between the person and the toy, while the autistic toddlers focused more on the toy. Forty of the young children involved in the study were diagnosed with ASD. With the help of the gaze, the researchers were also able to study how toddlers reacted to speech noises and observe early motor differences, since toddlers with ASD often exhibit postural fluctuations (a type of head movement).

Leveraging data science for the early detection of autism

Photo credit: Duke University

“The idea behind the app is to combine all of these behaviors to create a much more robust ASD algorithm. We believe that no function allows us to detect ASD in developing children because there are so many variations.” says Dr. Geraldine Dawson.

The app has multiple functions and enables ASD detection at home. Duke researchers are now one step away from starting a home study. Other benefits of this method include the ability to watch parents collect data once a month over time. In the future, this could be used in a treatment study to see if symptoms improve.

Duke’s ASD researchers are also working to incorporate information from the app into electronic health records (EHR) to see if information from routine medical care before the age of one can help detect it.

Children with autism and ADHD have more doctor and hospital visits in infancy. These are provided by the Duke University School of Nursing

Quote: Using Data Science for Early Detection of Autism (2021, April 5), accessed April 5, 2021 from

This document is subject to copyright. Except for fair trade for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.