Before letting the first patient take any medication we did a lot of testing to make sure the medication was both an accurate copy and also free of impurities. Unless the drug companies have faked the data it's reasonable to expect similar results.
We are tracking people in exactly the way you describe (with additional things like batch numbers and medication testing results) so with time will have some pretty valid stats on this, but at the moment we don't have enough data. Currently we have only seen 1 case where the patient has been non zero at week 4, but in a sample size of n < 100 the margin of error is +/- 10%
Real cure is when you not only get to zero, but some decent period after stopping the medication remain at zero.
Relapse represents not really being at zero, but rather being so low we can't measure at the end of treatment.
Most relapses will happen within 12 weeks of finishing treatment
cid.oxfordjournals.org/content/49/9/1397.full
We could set up an open data collection poll that people could use to add their personal data and then generate some graphs. While public crowd sourced data can be manipulated it would be interesting to see.