Detecting Health Conditions in Visual Content

I don’t watch mainstream news but once in a while I come across some and last piece I encountered was actually thought-provoking. Some viewers of a talk show hosting Denise Richards noticed she had enlarged thyroid gland and let her know that she should see a Dr. In this case, she had ignored the symptoms and looking at the photos, it’s quite obvious. But what about more subtle visual presentations of health conditions? Attentive doctors and specificalists may suspect or detect them watching a talk show, but that’s too few eyeballs. What if computers did the honors?

Google AI research has had years of history developing machine learning systems to detect various health conditions from medical imagery. I believe they started with retina scans looking for early signs of diabetes and more recently, they can detect skin conditions. Microsoft sports a similar history of research and publications. There is other research around detecting genetic conditions from portrait photos in combination with other data. I also remember some app which could detect cataract in photos with flash on. I’m sure much more has been explored in this area.

With the promise of ever improving machine learning in detecting patterns in imagery, specifically relating to health conditions, there are some interesting end-user product and service ideas that can potentially be game changing.

For instance, Google Photos or iCloud can run a bunch of more ML on user photos and videos and notify the user of neurological conditions, eye and skin conditions, strokes, mental health conditions, etc. Similarly, face unlock hardware such as those in iPhones and iPads could detect such conditions everytime the user looks at their phone (which is much more often than taking photos). And let’s not forget about the immense amount of public imagery content such as YouTube. Google already does copyright checks, transcription, thumbnail generation and a whole bunch of other processing. Disease detection will take it to a whole new level.

Apple has already invested quite a bit in health features of Apple Watch. It makes a lot of sense to expand that further.

Of course there are social aspects that I’ve not mentioned. Privacy, fake content, etc.