Why? Because the costs of public health mandate it. Costs of basic health services are already too high, never mind adding new services to the mix. Meanwhile, mining data that is out on the web, ready to be mined, is so cheap as to be essentially free (it has cost us no money so far to build SickCity, just a few dollars out of pocket).
So, in order to grow, become better, and not break the government bank, public health will need to start considering smart, cost-effective alternatives for services, and they will eventually look to using openly accessible data, data of all sorts - including social network data.
The SickCity team, btw, now has a statistician on board, looking over the algorithm to see how we can make the service more accurate. Version 1 works well in a very rough sense - how well can we make version 2 work, with someone who actually knows this stuff?