> Like with any medical test: if you do more tests you will find more incidence.

I've heard this line parroted a lot but a robust Bayesian analysis should mitigate this.

Like by setting appropriate thresholds you should be able to either

* do the same amount of interventions with more of them being helpful

* maintain the rate of helpfulness but doing more procedurees.

or even a mix of the two. Testing all the things all the time on all the people would be a boon if we just did it correctly.