Cancer "smelling" was presented on a recent TV show, "The Big C" in which a neighbor's dog could apparently able to smell Cancer. But, that is Hollywood too.
There was a recent discussion about smelling chips that could at least pick up some substances at very small concentrations. I didn't look at it closely enough to see how broad of a spectrum they were picking up.
I suppose the problem is defining exactly what you are smelling.
You already have various tools.
IR, NMR, PCR, etc... that can analyse various fluids & secretions for molecules or DNA. But, to a large extent they only would work with something that is well defined. I assume you could add sweat to the list of fluids to analyse.
An interesting would be to look for changes in a person's "smell" over time, not attributable to soaps & perfumes.
Maybe take 1000 smokers... and get a good whiff of their breath as a baseline, then test it repeatedly over the course of a decade. Smoke, of course, would confound the study, especially if they chose to quit (which you could not ethically discourage). Perhaps you could study ex-smokers who had previously smoked for an extended period.
You could also study other aspects of lung health such as pneumonia with a breath test.