UCI vs. Ashenden

February 14, 2013

This particular war of words makes for good reading. You really get an inside view into the process (or lack thereof) that governs the biological passport. It’s best to google it and read the whole exchange so you can see for yourself which side you choose to believe.

The brazenness of the UCI is staggering. With every strike they receive, they hit back without considering the wider implications of their statements. Take the assertion from the UCI that Lance’s profile was sent to the bio pass expert panel on May 4, 2009 and that the panel didn’t see anything suspicious. It further claims that the bio pass software didn’t pick up any anomalies after May 4, 2009 so that Armstrong’s values were never sent to the experts after that date. This is troubling for two reasons:

  1. Of course there is nothing suspicious in his tests up to May 4, 2009. That’s before the Giro had even started. Those tests couldn’t have been much more than establishing his bandwidth, so whether the values are regular or suspicious, you would have to be a complete idiot to violate the bio pass constraints that early on. In short; it proves nothing. If I, with my (according to the UCI) “very weak understanding of this very complex subject”, know this, then so does the UCI.To explain in very simplistic terms, the biological passport is an ever narrowing noose, as the test values that are picked up along the way create a profile of what is “normal” for this particular athlete. So if you’re tested the first time, a value for a human may be anywhere form 0 to 100. If you score 75 on that first test, then we can safely say you should never be below 50. So now your bandwidth for the second test goes from 50-100 instead of from 0-100 as it was for the first test. If that second test is 80, then we know the first test was not you are in fact in that 75-80 range. So for test 3, we may accept only a value from 65-90. And so on.
  2. The claim that the bio pass software didn’t pick up any anomalies after May 4, 2009 is actually very troubling. If that is true, and this is the software by which all samples are judged, then the whole bio pass system is a complete and utter waste because it obviously doesn’t pick up anything. It may explain why there are so few bio pass cases. Ignore whether you think the values from Lance’s 2009 Tour de France prove he doped. What everybody – believers and non-believers – should be able to agree on is that the values are at least suspicious. Or if you insist on an even more benign word: atypical. And the process should be that the software picks out profiles that are suspicious, or atypical, so that the committee can analyze them further. There’s lot to dislike about that process even if it does work properly, but the UCI seems to now state that suspicious profiles don’t even make it to the committee because the software fails to detect them. So in a panicked attempt to explain away their inexplicables, they are now throwing their own biological passport under the USPS team bus.

In the end, there are only two possible conclusions:

  1. Armstrong’s profile was flagged by the software, a sign the tool may be useful, but the UCI decided not to forward the profile to the expert panel for review.
  2. Armstrong’s profile was not flagged by the software, which means the software and therefore the whole protocol is useless.

For the people involved it matters little. Either they knowingly withheld a suspicious profile or they have lied about the effectiveness of the biological passport process they use (you remember, the process that proves they are doing a good job and that cycling is getting cleaner). Either way the blame lays with them.

Expect two things from the UCI in the next few days:

  1. McQuaid will make his trademark switch from “I’m in control, trust me, it’s getting better” to “It’s not my area of expertise, you should ask the experts, the passport is actually the work of WADA”.
  2. Silence

33 Responses to “UCI vs. Ashenden”

  1. Ashenden himself has answered this question: LA’s anomalies were such that the software wouldn’t have picked them up, needed a human read to be spotted.

    • Luis Oliveira Says:

      That’s nonsense! What Ashenden said was that “the software was not engineered to detect that abnormality”, not that “the software COULD NOT detect that abnormality.”

      Just design a proper software (or continuously evolve to current one. Quite simple.

    • As I said, if that’s true, then the software sucks. We have a profile which at least one expert says only leaves a 1 in a million chance of not being that of a doper.

      Hyperbole aside, that should mean there is at least a strong possibility. If the software doesn’t pick that up, then obviously using software to pick which profiles the experts should see is a stupid idea.

      • This is the point I keep coming back to: at what point do you move beyond software prompting a manual review and actually rely on human intelligence?

        Obviously you cannot have the experts review every single passport (1,100 in the RTP, and there’s supposed to be an average of 10-12 tests per rider per year, there’s simply not enough time). So you have to be selective and have rules for triggering manual review.

        We may assume that Sotas had a good reason for not designing the software to catch anomalies of LA’s type, or we may assume that there was a flaw in his thinking here (even allowing for the fact that WADA have signed off on it) but right now we need to park that and deal with the way the system is actually designed and actually works.

        In other words, is there any reason why a targetted, intel-based manual review could not have been triggered? Is there any reason CAS would rule this wrong?

        • Actually, it’s not that much work. 1,100 profiles (the number of tests in the profile isn’t that important from a review point of view, although it’s much less than 10-12/year nowadays). So that’s 5 profiles per work day if you want to review them once a year. That’s basically 1 profile per expert to review and flag if it should be looked at by the complete panel or not.

  2. Ate Says:

    What I understood about this is the following: The software can pick up a deviation from the normal values. However with Armstrong’s case it didn’t pick up the fact that there should have been a deviation(because of riding the TdF). IF the software is not designed to know about a rider’s expected performance due to racing it should be changed.

    I believe it should be possible to implement an algorithm that picks up strange (non)deviations as well as human tester’s should. There is more than enough literature in Machine Learning research about outlier detecting etc. (often used to automatically flag credit card fraud)

    The biggest hurdle faced in such software is the pipeline through which values are determined, entered in the software and processed, the actual detection algorithm is probably a small part of this program, and if the software was designed with future adaptions in mind it should very well be possible to change and improve these detection algorithms. Whether the UCI wants this is another issue. In any case, the head of the UCI should be held responsible for grave cases of cheating and the possibility to do so.

  3. flammecast Says:

    Actually there’s a third conclusion – which is even more damaging to the sport – Armstrong’s Profile was flagged and the software parameters for flagging a profile were reset so that the profile was no longer flagged, rendering the process null and void..

  4. djconnel Says:

    The claim the software found no anomalies is obviously true only if the software was never given the values to analyze. I plotted Armstrong’s self-published numbers here. The points go from one corner of the plot to the other. It couldn’t be clearer something changed.

  5. Daniel Says:

    Is it as simple as staying below a 130 OFF score?

  6. RabAusten Says:

    Which is why astronomers often use crowd sourcing to identify oddities, like new planets or galaxies, because software cannot always be made to pick up subtleties in complex data/images. Whereas the human eye/brain can see intuitive patterns or anomalies.

    Either way though the egg is firmly flying in the direction of the UCI’s face.

    • atejf Says:

      This is not entirely true RabAusten, Machine Learning and Data Mining is becoming extremely powerful, but more as an addition to human expertise than a replacement. However detecting ‘abnormal’ or ‘atypical’ blood values shouldn’t be as hard as detecting new galaxies, stars and planets.

      If the expertise put into the software is incapable or the data is insufficient(not just the blood values but also other circumstances) the software will also be incapable and insufficient.

      • I agree, just pointing out the analogy of Ashenden’s statement as given by @fmk_roi to other real world situations. Maybe UCI has been sold a pup by the software company or maybe nefarious intervention/obfuscation.

  7. Rob Churchill Says:

    Ashenden says anomalies of this sort aren’t flagged by the software but are easily spotted by the experts. Should be easy to improve the software then. In the meantime, all stage winners, podium finishers, jersey winners etc should automatically be appraised by the human panel, along with a random selection of others. Simple.

  8. @RabAusten

    Yes crowd sourcing suspicious profiles would be awesome.

    Only caveat is I doubt is. Is there enough data to keep the “crowd” occupied. Many you could get the pros to do it?

    • Not sure letting the public loose on this data is the answer though.

      • I don’t think it is. Publishing this data in general is not a fantastic idea i think. Of course there could be open and sut cases of doping or clean riders, but there will be a lot of stuff in the middle, things that look suspicous at first glance but aren’t if you know the race profile, weather conditions, illnesses, etc. That subtlety will be lost on the general viewer, making pretty much everybody a suspect in the public eye.

        • hughw Says:

          Pretty much everybody already is a suspect. Public access to the data won’t worsen that, and might improve it.

  9. Question that should be asked ?

    When will phat mc splat apologise to you ?

    Answer ! Silence ß

  10. Evan Shaw Says:

    The real issue is whether the UCI wants to know or not know. And the evidence is consistent with they do not want to know when it serves them to cover up. Those who have nothing to hide hide nothing.

    Gerard, please see this scientific video explaining exactly how we know that Armstrong was indeed doping during the TDF comeback.


    The reasoning synopsis is that blood volume non doping decreases during a long tour, and bone marrow production increases. This is what we see in LA pre Giro and Giro. But after Giro, he goes to altitude to provide a cover for massive increase in hematocrit and then during the TDF NO blood volume decrease and marked bone marrow reduction. But he takes micro doses of EPO to keep bone marrow off of the floor to avoid triggering the passport.

    This is highly valid scientific proof of sophisticated doping and the Astana doctor they had clearly managed and regulated these parameters to provide LA and perhaps others with this advanced evasion methods.

    Clearly the UCI is making it look like a war of words but just as LA is not a bully but a sociopathic criminal, the UCI is using the very same tactics of attack, destroy, and invalidate all who threaten to make them accountable.

    This is serious business. He was not just a bully and the UCI is engaged in a systematic cover up.

    It will not be enough to change the culture of cycling. We must reinvent the UCI.

    Armstrong used the big lie told over and over. So does the propaganda and demagoguery methods of UCI.

    Enough is enough.

    • Yes, I’m familiar with the video. It’s a good primer for people. I hope it also shows it’s not as easy as simply looking at peaks or valleys in a graph.

  11. Having seen the UCI activities since 1960’s it is probable that the best available data is correctly interpreted and “It’s not my area of expertise” would be the applicable reply. Since an executive decision must be made “as the ruling Authority we’ll do our own thing without justification. It worked OK when we “dropped Katusha using our andecalered Ethical excuse”! Worked then without problem so it will work for the biological passport so stop querying our methods.

  12. larryatcycleitalia Says:

    This is all just the same old s__t from these crooks. Say something, anything that might push the scandal of the moment out of the public eye, whether it makes any sense or not. It’s clear that all the detection methods the UCI has any control over are simply wallpaper, just as the 50% hematocrit limit was – “make the general public think we’re doing something, while we’re actually doing business as usual” That’s why they hate WADA so much, they don’t have much control there and those guys (plus USADA) actually try to catch cheaters rather than manage scandals. Cycling needs an entirely new governing body – who will step up to provide it? Is there anyone who can put the long-term interest of the SPORT first?

  13. Justin Says:

    All of this talk about software is interesting, but misses the main point as far as Armstrong and the appearance of UCI corruptioin is concerned. The purpose of the software is to flag “suspicious” passport values so that they can be referred to experts. In Armstrong’s case the UCI did not need software to flag his values, indeed, his values was in the public realm, experts (including Dr. Ashenden himself) had stated that the values were suspicious, and the UCI was aware of this fact. So whatever some computer in Lausanne thought about Armstrong’s values, the UCI was actually aware that they were suspicious. A summary is here:
    How is it that a computer trumps the UCI’s actual awareness that cheating is happening?

    Gerardvroomen’s wider point about the software possibly being configured to render the protocol useless is valid.

    • Rod Says:

      I would posit an even more perverse theory. Some of the testimonies regarding Armstrong’s “donation” (donations?) stated that it was explained to him how the tests worked – which would unlock how to fool them.

      Perhaps this not only included the EPO blood tests, but the threshold values, algorithms, etc. for the biopassport. If LA could stay below the “auto-flagging” values somehow the UCI could claim that the system didn’t point them out and therefore they would not require being sent to the experts for visual assessment. Therefore, no results would be sent – coincidentally, those that correspond to TdF preparation and race.

      And if the crap hits the fan, the UCI can always say ” this is the way the system works”. Even if it is designed to selectively fail.

      Needless to say, after all the backtracking by the UCI regarding Armstrong, biopassport, WADA code enforcement, etc. I have zero confidence in them.

  14. Evan Shaw Says:

    It is NOT rocket science to know how to look at these tests. Courage is the willingness to look at the obvious. Watch the video. My 11 year old son can understand it.

    • Rocket science is simple. But while your 11-year-old may understand the video, I’m not sure he understands the intricacies. Just to contemplate, I have spoken to experts in the field who agree with the gist of the video but not with some of the detail statements in it. The interpretations are open to, well, interpretation.

  15. BC Says:

    Don’t get to bogged down in the minutia. We are dealing here with the UCI. Until such time that the whole rotten lot are replaced with something transparent and independent there will be no change. Oh, Schleck to CAS anyone at UCI HQ ?

  16. Drew Says:

    I still say… Lance had the programming of the software contracted out. It was programmed to automatically adjust the data in his favor. His values were anomalies that scientist typically remove from the data to make the sampled population conform to what they expected to see. I have seen this done many times in research and the Bio Passport system is no different in the process: take a bunch of samples; plot them; drop the samples that are outside the norm. Done! Looks good! Nothing to see here!

  17. Skater Says:

    Why is there software at all? If there would be millions of blood tests…but noooo. A few here and there. Let an expert look at it. He need nog not the name of the athlete. Just the timings and nature of races undertaken, which races were DNF’s, etc. The expert can also ask for related urine sample timings and request specific re-tests. Like EPO for microdosing. Or, if peritted, blood for plasticizers. Etc, etc.

  18. I’m closing the comments on this page because for some reason it is targeted by spam porn comments and I’m getting tired of having to remove them all the time. Sorry for the inconvenience.

Comments are closed.

%d bloggers like this: