Hi,
I was expecting to use the threshold probability of 3% in this question. It seems to me that miscalibration would be most concerning around that probability but the risk group numbers we are given do not support this. Maybe I'm missing something?
Thank you in advance.
What you are missing is the "if any" in the question.
In
which risk category (combination of Wells group and D-Dimer interval) if any would
miscalibration present a significant problem?
I usually avoid this kind of trick. Sorry.
Michael
Why not be even more clear? Disagreement between predicted risk and observed proportion with the outcome (i.e. mis-calibration) is not a problem as long as both the predicted risk and observed proportion are on the same side of the decision threshold.
Yes, that was what I thought too. Thank you.
