Diagnosing diabetic retinopathy requires a skilled physician to analyze photographs of retinas. Early diagnosis enables about 90% of new cases to be controlled with proper treatment and eye monitoring.
deepsense.ai’s model reduced the time and effort needed to identify developing diabetic retinopathy.
A weighted Kappa measure based on the opinions of trained physicians was used to validate the solution. The model scored 83% QWK, a result comparable with what a skilled physician can do.