Ok but I guess the question is, is 10% better or worse than the diagnosis rate of a group of specialists?
Also, I imagine there are massive benefits if people all around the world were pumping data into a system. Correlations that a group of specialists may not think to look for may start to appear.
For lyme specifically, the blood tests are generally inconclusive, symptoms vary from person to person and are inconsistent, and even doctors who specialize in lyme seem to disagree on core details of the diseases' effects.
If consensus can't be reached even at that level, and if that level of complexity is present in many other ailments (which I assume it is), how would be able to build a beneficial piece of software around that information?