Differential diagnosis is more fit for computers than surgery, and yet all you read about is robot surgeons.

This. Most differential diagnosis is so computed operation like, that I am amazed how little people use it. I really wonder how come in every doctor’s office there isn’t already a software to provide suggestions based on symptoms and then learn from medical diagnosis and the outcomes of the symptoms.

Neural networks could analyze even more data than doctors, simply because you cannot remember to tell your doctor everything as well as a computer could: for example in every grocery store what you buy, how much you buy, how often you buy are all great indicators for nutrition, body chemical balance and the gradient of related affections.

All the new Health Kits, the intelligent things (for example if you live alone or in a simple couple, Nest can tell if you are suddenly setting abnormal temperatures in your house), even wishlists can feed data into a predictive neural network informing your doctors of potential paths to follow for your diagnosis.

And what’s even more interesting is that, as I said before, the cloud keeps getting better as soon as it finds out what the disease actually was. There may be businesses doing this already, I didn’t research this before responding. But nevertheless I still wonder why aren’t we having more of it yet.