This is precisely the problem facing the medical sector at the moment. The number of people complaining of depressive illness in the last 10 years has gone up something like 300%. They can't all have depression, but so far there is no definitive way to tell if someone has a neurological disorder. Even if you did a spinal tap and showed someone's central serotonin levels were a little low is inconclusive evidence - serotonin levels naturally fluctuate to some extent, and an abnormally low reading could be anomalous.

A good way to tell if someone really is suffering from the condition is something called a Beck Depression Inventory. Google it. People who are fine score under 10. People who are just a bit sad tend to score 10-20. Those with a serious illness rarely score below 30 and those above 40 are a suicide risk. At my worst I scored a 38 and at my best a 0. My dissertation was, in a very contrived way, looking at ways of actually clinically diagnosing depression. We had a look at how taste changes when serotonin is manipulated, but this stuff is in its very early stages. Even if it eventually works (which I'm skeptical about) it'll be decades before there's a conclusive test for depression.