AI model identifies pediatric eye diseases using cell phone photos
Click Here to Manage Email Alerts
Key takeaways:
- Researchers found that an AI model could identify eye diseases in children using only cell phone photos.
- The study was conducted in China, where around half of children are affected by myopia.
An AI model was able to identify several common eye diseases in children using cell phone photos that could be taken at home, researchers reported in JAMA Pediatrics.
The study was conducted in China, where around half of children have myopia, or nearsightedness, “the most common eye disease among children and teenagers,” according to Qin Shu, MD, a physician at Shanghai Ninth People’s Hospital, and colleagues.
“Myopia ... has increased across numerous countries and areas worldwide, emerging as a global public health concern,” they wrote.
The study also tested the AI model’s ability to diagnose two other eye diseases — strabismus and ptosis — that are less common among children in China.
For the study, 476 children who had already been diagnosed with at least one of the three eye diseases at Shanghai Ninth People’s Hospital provided 1,419 photos of their eyes using smartphones.
The photos were taken in the hospital, cropped, then fed into a deep learning network to build the AI model. Afterward, the researchers tested the model’s ability to correctly identify the three diseases using only the cell phone photos.
The model accurately identified all three diseases but performed best in the diagnosis of ptosis. Among the metrics the researchers used to assess its performance, the model demonstrated a sensitivity of 0.85 (95% CI, 0.82-0.87), specificity of 0.95 (95% CI, 0.93-0.97) and accuracy of 0.92 (95% CI, 0.91-0.93) for ptosis.
The scores were lower for the other two diseases but still significant, according to the researchers.
Although the photos used in the study were taken in the hospital, the researchers said the objective is to provide a way to “facilitate the early detection of pediatric eye diseases in a convenient manner at home.”
“Moreover,” they wrote, “using such information can help achieve a more equitable allocation of limited medical resources. This is critical to the advancement of global health standards.”
Collapse
Read more about
Click Here to Manage Email Alerts