File photo shows babies at a baby race in Xiangyang, central China's Hubei Province. (Xinhua/Yang Dong)
The system can be used for the screening of infant visual functions as well as the timely detection of infant visual impairments.
BEIJING, Oct. 24 (Xinhua) -- Chinese researchers have developed an intelligent system for evaluating infant vision with deep learning technology, according to the China Science Daily Thursday.
Researchers from Zhongshan Ophthalmic Center of Sun Yat-sen University developed the system based on the behavioral changes of visually impaired infants. The system can be used for the screening of infant visual functions as well as the timely detection of infant visual impairments.
Researchers video-recorded the behavioral patterns of 4,196 infants who experienced varying degrees of visual loss and developed a deep learning algorithm trained with the videos to find out how vision loss is associated with behavioral changes.
They identified a quantitative relationship between the frequency of more than 10 behavioral patterns and the severity of visual impairment.
The results showed that video recording of behavioral patterns can help evaluate infants' visual function. Compared with traditional detection methods, the system has higher feasibility and accuracy, which has less demand for infant cooperation.
The research also offers a reference for further exploration of the visual development of infants and young children, according to the report.
The research was published in the journal Nature Biomedical Engineering. ■
