AI-Powered Autism Screening: Towards Early Detection and Reduced Disparities

AI-Powered Autism Screening: Towards Early Detection and Reduced Disparities
Scientists are using AI for autism screening and reduce healthcare disparities. (Image: mikimad/iStock)

Leveraging AI for Objective Autism Screening

Duke University researchers have been investigating the potential of artificial intelligence (AI) for screening autism in real-world scenarios, specifically in primary care settings. The aim is to create a reliable and scalable tool for early autism detection that does not solely rely on parental input. Current guidelines from the American Academy of Pediatrics recommend autism screening for all children during their 18- and 24-month well-child checkups. Traditionally, this involves a 20-question survey completed by parents, but this method has limitations, such as language barriers, understanding issues, and a potential lack of follow-up interviews by pediatricians.

In contrast to traditional screening methods, other medical fields utilize multiple sources of information to assess the likelihood of a medical condition. For instance, if someone is concerned about a heart problem, a doctor would inquire about symptoms and conduct various objective tests like an electrocardiogram (EKG) and blood pressure test. However, autism lacks objective tests, and its diagnosis relies on behavioral observations. The Duke University research team has addressed this gap by developing a digital app for smartphones and tablets that utilizes AI to screen for autism in a quick and objective manner.

Digital Phenotyping with High Accuracy and Equity

The app functions by displaying short, engaging movies strategically designed to elicit behaviors associated with autism, such as gaze, response to a name call, and facial expressions. As the child watches, the device’s camera records their behavioral responses. Computer vision analysis is then employed to precisely measure these responses, including attention to social or nonsocial elements, facial expressions, blink rate, and body movements. Machine learning processes these signals to determine the likelihood of autism, all within a time frame of less than 10 minutes.

One notable advantage of this “digital phenotyping” approach is its high resolution and accuracy, enabling the detection of subtle behaviors that may go unnoticed by the human eye. For instance, the computer can detect differences in blink rates and the speed of head-turning, which are early signs of autism. In a recent study involving 475 children during a well-child visit, the app demonstrated 87.8% sensitivity in detecting autism and 80.8% specificity for those without autism. Importantly, the app proved equally accurate across children of different ethnicities and genders, potentially addressing disparities in early detection.

This innovative use of AI in healthcare, specifically for autism screening, highlights the technology’s potential to transform medical practices. While AI cannot replace the human touch in healthcare, its ethical and responsible application offers numerous benefits, including increased access to services and more efficient and equitable delivery of healthcare.

Google News Icon

Add Slash Insider to your Google News Feed

Source(s): Psychology Today

The information above is curated from reliable sources, modified for clarity. Slash Insider is not responsible for its completeness or accuracy. Please refer to the original source for the full article. Views expressed are solely those of the original authors and not necessarily of Slash Insider. We strive to deliver reliable articles but encourage readers to verify details independently.