AI-based early detection of visual impairment on smartphone: study

A child takes a visual impairment test on a smartphone. (Photo: CGTN)

Untrained people can use smartphones to detect visual impairment among kids with AI technology, according to a study by scientists from Sun Yat-sen University and Shanghai Jiao Tong University.

Visual impairment is one of the most important causes of long-term disability in children worldwide. Still, its early detection is frequently missed since kids have limited cooperation with standard vision tests.

The study published in the journal Nature Medicine on Thursday described a smartphone-based system, the Apollo Infant Sight (AIS).

The AI-driven system is shown to identify visually impaired children with 16 eye disorders by recording and analyzing their gazing behaviors and facial features.

The AIS is designed to attract a steady gaze from children by playing cartoon-like video stimuli and then capture the facial appearance and ocular movement for further analysis using deep learning models, according to the study.

A smartphone screenshot showing a cartoon-like video and a child watching it. (Photo: CGTN)

Videos from 3,652 children under four years old have been prospectively collected to develop and validate this system.

The study shows that untrained parents or caregivers who use the system on their smartphones are able to achieve high-quality detection.