You can't patent the fact that a cotton gin works, nor can you patent the use of a cotton gin. You can only patent the design of the gin itself.

With drugs, you should only be able to patent the design of the drug itself, and probably also the method of synthesizing the drug. You should not be able to patent the idea of using aspirin to treat headaches.

Your patent on a cotton gin is on the design, not the use - you don't reap patent fees every time your dickensian worker operates it.

The difference is that one is knowledge, and another is a mechanism. Someone has to build and produce a cotton gin, to infringe on a patent. On the contrary one only has to learn about the correlation and now doing anything with this knowledge is "patent infringement" ?