An app based on artificial intelligence has been successfully developed and tested - Apollo Infant Sight (AIS) - which helps doctors and parents to diagnose vision disorders early in small children even under 2 years of age, accurately distinguishing between 16 different ophthalmic disorders only through the study of the gaze of the child.
Reported in Nature Medicine, the study is due to Chinese researchers at Sun Yat-sen University.
Experts recorded the videos of 3,652 Chinese children aged 48 months to develop the app.
They have validated this system twice in a clinical setting and also at home as a DIY.
AIS captured stills of children watching cartoon-like videos to identify vision problems.
He accurately distinguished 16 ophthalmic disorders in children by analyzing their facial features, gaze behavior, and eye movements during visual stimuli.
Young children, especially newborns, are unable to undergo standard vision tests, making it difficult to detect any impairment early.
There is an urgent need for easy-to-use tools for detecting vision problems in young children.
The researchers collected over 25,000,000 video frames of the 3,652 children using AIS.
Validation of the system took place in three phases, in five clinics and at home.
In the first two phases, children enrolled in the study underwent ophthalmic examinations by clinical staff and trained volunteers collected videos.
In the third phase of the study, the experts recruited the parents of a group of children online to validate the use of the app at home.
The parents, although not experts, recorded the children's videos with their smartphones following the instructions of the AIS app.
All children underwent ophthalmic examinations at a clinic.
AIS has achieved high values of sensitivity and specificity both when used by ophthalmologists and when used by parents.
The app could help in timely referral of visually impaired children to pediatric ophthalmologists, who could start treatment on time.