Specificity: Calibrating the BMF
A lot has been made of error bars (or the lack of on the DpVAM and DpW/kg analysis)
Thinking in terms of error bars is a bit of a lab based thing (ensue the rebuttals from “real” scientists)
The DpVAM and DpW/kg are meant more to work like a medical test. (Just think of the mass confusion of a pregnancy test with error bars)
In this context, doc proposed a simple interpretation 2 bars up = likely doping
split bars = suspicious
Up to this point doc felt that Specificity of the test was probably good
especially when considered over the course of a Grand Tour.
Specificity, is the likely hood that a positive result is a true positive
2 bars up really = doping
To test the Specificity a Gold Standard Clean Rider was needed.
The best Gold Standard of a rider considered to be one of the best ever
widely regarded as completely clean
So doc used Vayer’s data for Lemond in Not Normal.
To also, add a modern GT winner doc added Ryder Hesjedal.
Although, many may not consider Ryder as a truly top GC rider
he is the only GT winner with published blood data
that is not suspicious for doping.
(doc judged his chronically elevated reticulocyte count is likely his natural baseline as the level of EPO needed to achieve these counts should have tested positive by now)
So with these “Gold Standard Clean Riders”
doc tested the physiologically based DpW/kg
(pW/kg is based on a VO2max of 90 ml/kg/min with a 23% efficiency)
(DpW/kg is this same rider doped with the equivalent of 500 ml of fresh blood ie 1 transfusion)
so with the proposed thresholds
the new phsyiology based DpW/kg
has 3 false positives
0 false positives for entire Grand Tours
for a Specificity for doping of
on any given climb
for Grand Tours
if yer comfort level is a bit more pragmatic
using split bars for suspicion
produces 7 false positives
out of 22 on any given climb
0 false positives
again 99 % Specific for doping
for Grand Tours
fully acknowledging the limitted data set used as the “Gold Standard”
this (preliminary) result