It doesn't have to show that they will get Parkinson's 100%, just that they progressing towards it and would very likely get it if they don't change their lifestyle and lived long enough.
It is not a test of their genome - it needs to be of some active biological state of their body.
Current theories suggest people who get diagnosed with Parkinson's have obvious (in retrospect) symptoms throughout their body, and that these are likely diagnosable chemically, even though actual diagnosis typically only happens when the patient reports the typical symptoms: one-sided stiffness, slowness of movement, etc.
https://www.youtube.com/watch?v=uTB5MmQIIyM SYN-ONE skin biopsy test seems relevant. They are a bit light on explaining how predictive it is (for not-yet diagnosed people) but they claim over 95% sensitivity (at least for the misfolded AS; not sure of the correlation rate of that to actual PD)