It would seem that clinical diagnosis by machine is no longer a far-off vision of future health care, with the recent announcement that the US Food and Drug Administration (FDA) has approved marketing of artificial intelligence (AI) to detect a form of eye disease without the assistance of an eye specialist. IDx-DR is software designed to detect Diabetic Retinopathy by the use of image interpretation. Images of a patient’s retina are taken with a special camera, and then uploaded to a cloud server, where they are evaluated using an algorithm. In trials involving 900 patients, the software accurately detected more than mild diabetic retinopathy 87.4% of the time, and less than mild diabetic retinopathy 89.5% of the time.

IDx-DR is one of several recent AI packages which are being taught how to diagnose disease. Earlier this year in the UK, the NHS announced a collaboration with Google’s DeepMind to use “deep learning” in order to diagnose eye disease (see our blog https://www.seriousreaders.com/blog/nhs-collaborate-with-google-deepmind-to-fight-blindness).

However, using AI for diagnosis without specialist intervention is not without its problems, not least being the issue of responsibility. Who ultimately is to blame for a wrong diagnosis? One suspects such issues may just be resolved in various courtrooms over a period of time.

Sources:

1. FDA:
   FDA permits marketing of artificial intelligence-based device to detect certain diabetes-related eye problems.

2. Science News:
   In the future, an AI may diagnose eye problems.