VR3 BUH or VPM

Dutchy

It's called erring on the side of caution????

__________________
= This post is environmentally friendly. It is composed of 100% recycled electrons only. =

aainslie

The best way to do that is to not do deco dives.

And I have no problem with erring on the side of caution. I have a problem with using different rules for using data depending on your mood. It's just poor empiricism.

__________________
Andrew Ainslie

AM

Just because it has been framed in the originating perspective of "I know of..X, Y and Z" does not negate the fact that there IS empirical data out there about the long term effects of hyperbaric exposure. If you are going to compare anecdotal evidence and say it is not acceptable versus "hard data" gathered against V Planner dives, then show us the hard test data for V Planner. I am seeing here a considerable propensity for "inferences" as to how the VPM model has been tested, as well as "man hour" statistics. Any proper academic study against VPM model would incorporate control subjects and would factor in the complex physiological variables that exist in decompression diving, as well as encompassing PURE VPM diving, without padding which takes it to the level of bare ZHL-16 diving.

Practically every deep diver I know in the UK who used early versions of VPlanner was bent beyond 100m (and a lot shallower/shorter in some cases) and accordingly the model was revised. This constitutes the use of anecdotal evidence to make substantive differences to an algorithm, the very practice you suggest we should not encourage.

The model therefore at that stage, was not revised according to a scientific subset of principles (relating to the mechanics of the algorithm) but according to the physiological experiences of a group of divers. This is not empirical testing, anyone who thinks this approach constitutes formal testing needs to go and think about the implications of using an aggressive model being revised against actual DCI reports. The more sensible start point would be to start from a position of conservatism and work back slowly, according to the principles of the algorithm.

There are several adopted practices used in conjunction with VPM diving, some of these practices serve to significantly pad out the bare VPM model, therefore any conclusions drawn from these such dives and re-labelled as "test data" in fact, constitute completely erroneous data points, serving as another example of the "anecdotal" evidence we are told is not acceptable in the "rules" of data analysis!!

Again, what academic, scientific, statistical, valid testing methodology has been instituted to collate data determining the efficacy or otherwise of VPM?

As well as anecdotal evidence about residual long term damaging effects of poor decompression, there are *multiple* academic sources of this info, collected within the parameters of a professional scientific test approach. Maybe when I have more time I will dig them out, otherwise you'll have to do some research. It's not hard to find this stuff, use the commercial sector as a start point, even 50m air dives with "adequate" decompression are shown, by medical baselining, to cause long term residual damage!!

Regards

AnneMarie

__________________

aainslie

Anne Marie,

let me phrase this very simply.

1) Used correctly, I think that anecdotal (or in general, qualitative) data are useful

2) A large number of dives with low incidence of issues (essentially an uncontrolled experiment) is somewhat more useful.

3) Clearly, controlled experiment data are extremely useful, and the best of all. Although it IS important that one is clear about the hypothesis being tested, and that one only use the results in taht context. For example, as Ross poitned out, the hypothesis being tested in the French experiment is essentially that EXCESSIVE DEEP STOPS WAY LONGER THAN ANY MODEL RECOMMENDS are bad. D'uh! It CANNOT be interpreted to mean that VPM, RGBM or GF are bad since it replicates none of these. yet people walk away making general statements about deep stops ALWAYS being bad. This is poor interpretation of the data.

In eariler posts Mike claimed that type (2) was not useful. Then he went on to use type (1). That's inconsistent.

That was my only point.

The inference I was hoping people whoud draw is, Mike should accept the (lower, but still acceptable) utility of data of type (2) before he starts relying on type (1). However, to be clear, let me state that inference here.

And once more, to be totally clear, I have no problem with ANY of these types of data, as long as they're used with appropriate caveats.

I made NO statements about actual data sources in that post about Mike's use of anecdote. Only about methodology of interpreting those sources.

Thi is pretty simple stuff and anyone with a reasonable scientific education can appreciate this.

__________________
Andrew Ainslie

AM

You can phrase it any way you like Aainslie but that does not alter the material fact that you are defending the indefensible. What type of anecdotal data is useful? Truthful, accurate and honest anecdotal is possibly useful. Positioning a website to demonstrate dives on VPM that we KNOW categorically were NOT conducted on VPM is NOT useful?. That is not qualitative data, I think the word you are forgetting to use in accurately describing this practice is LIES. There is NOTHING anecdotal about altering the evidential circumstances of a dive to create a position of support for an algorithm which WAS NOT EVEN USED ON SAID DIVES.

No, once again, this is just the utilisation of statistics to create false platitudes of model integrity. The VPM model is not empirically validated by the reporting of a range of shallow dives with little or no deco. There is an directly proportional relationship between increasing depth/time and incidences of DCI. So any statistics that you produce showing VPM usage in increasing depth ranges would support the proposition that DCI incidence will INCREASE with depth, not remain low in proportion to the number of dives conducted. If there was true confidence in this ridiculous test model and these nonsensical propositions, there would be no problem with publishing this so-called "bend and revise" test data. Every diver that I know that has used VPM to any reasonable depth has been bent.

No, the hypothesis is that excessive deep stops may be bad if not adequately combatted by sufficient shallow decompression. Deep stops on their own are of course not good if their impact upon slow tissues is not factored into the whole decompression plan. The enshrining and manifest principles of deep stops are bound in the intrinsic function that they perform, that of offgassing from fast tissues whilst ongassing to slow tissues takes place. The fast tissue offgassing requirement is a function of the fact that these compartments have been proven to reach saturation quicker than slow compartments. If sufficient shallow decompression is not undertaken after deep stops, then the loading that has occurred into shallow tissues (during deep stops) will result in Type 1 DCI. You are getting confused about the general statements people may make about deep stops. They are only bad when not used in conjunction with sufficient shallow decompression and this is a fundamental problem with the VPM model and it's variants.

No, what is inconsistent is producing data from dives that were not even conducted on the VPM model and then using that data to infer validation of the model. Even if that data had been collected from VPM use, it uses man hours (1 in your list) and then also falls under the banner of 2) in your list by suggesting a high number of dives with low incidence of DCI, with no evidence to support the proposition that increasing depth relates to increasing DCI incidences. Therefore by your own admission the data that is being supposedly used to provide validation of the VPM model, is neither accurate, correct, useful nor consistent.

No, none of us should accept data from your group 2, which on a pure statistical, scientific and probability basis, negates a fundamental principle in decompression diving, the ACTUAL relationship dynamics, i.e. a directly proportional correlation, inherently found between increasing depth and increased DCI incidents. We could take a large group of shallow dives with no incidents and draw the inference that this validates the model but this does not provide ANY meaningful data other than the model is comparative in this data range, with other models. It does not validate the efficacy of the VPM model for DEEP diving nor does it provide any valid comparison mechanism for other algorithms used in DEEP diving.

I have a major problem with the types of data methodologies that you propose as being sufficient to validate an algorithm, caveats or not. The approach taken to date, in revising the model *backwards* is patently unacceptable, as is misreporting the dive data.

Your data sources have been irrefutably proven to be mistaken and misleading. The methodology of interpreting those sources is also misleading because it does not compare a range and incidence data point alongside the very algorithms you claim are less valid.

What anyone with a reasonable scientific education can appreciate is the latent misleading data being presented to infer validity of the VPM model. If you disagree and no doubt you will, then publish the raw data used to "test" and "validate" the VPM model and let users decide for themselves, rather than presenting the product with your own incorrect and false conclusions.

__________________

aainslie

Fine. Tell Mike that you agree with me and that he shouldn't use anecdotal data. At the core, that was my first point.

And might I point out that you are now contradicting yourself? In post 122 para 3 you condone using anecdotal evidence (actually, it's really uncontrolled experiment data but let's not nitpick) then in para 1 post 124, and again in para 2, you claim you CAN'T!

So, which one is it? Should VPM have been updated based on the data coming in no matter its source, or should they have gone into ivory tower mode and said that nothing should be done until a controlled experiment was run? I for one am glad that they were sensible about this.

There's a hierarchy. It's silly to claim either that all are equally useful, OR that only one is acceptable.

__________________
Andrew Ainslie

AM

Absolutely not. Not a hope in hell. I don't and will never agree with your data methodologies.

There is a big difference between anecdotal data (backed by substantiated scientific data) and blatant LIES propagated to create perceived validation of an algorithm. You, unfortunately, fall into the latter camp. Your real issue is that if you used true anecdotal data (with the DCI data attached), it would reveal multiple bends and be devastating to your commercial interests, therefore you have developed a latent bias against real and honest anecdotal data. Furthermore, you select data from a relatively fixed range of depths/times to infer that the model can support lots of dives with few incidents. A bullshit proposition if I ever heard one.

If Kevin Gurr does not respond to your internet claims about VR3 Buhlmann, that does not, contrary to your assertions, automatically signal his "acquiescence" to your claims about the VR3.

Please publish your test methodologies and full data for the VPM model before you make the assertion that it is "better" than any other model.

To Lance - your Buh VR3 is showing you an extra hour of deco for a very good reason. You'll most likely get ****ed on a deeper/longer dive if you trust VPM above Buh. Please think carefully about electing to cut the deco massively short (compared to Buh) using the VPM model, which has built it's very foundations on revisions following DCI reports and misreporting of data sources and methodologies.

Regards

AnneMarie

__________________

aainslie

Good grief, to quote Charlie Brown!

What evidence do you have that Mike's anecdotes are correct and somenoe else's (I don't even know what your caps are all about) are lies?

What do you think my "commercial interests" are? I'm a tenured professor, and have no commercial interests in ANY scuba device.

How do YOU suggest we resolve the issue of what Gurr has implemented, since he refuses to divulge it?

I don't have any data on any model. I was talking broadly about research methodologies. it's something I know a bit about. And I didn't assert that any model was any better than any other. I happen to think that VPM makes a lot of sense and I'm a big fan, but that's about it - I have no proof that one or the other is better.Actually I don't really think there's any compelling argument anywhere that any model is better than any other. Especially since VERY little analysis discusses what really matters - the benefits of an earlier exit from the water vs the higher risk of DCS, and which (if any) model does the best job of balancing them.

A fun scientific study (as long as I'm not the guinea pig) would be to take a GF and a VPM model, a dive to a known depth (say 200 ft for 20 mins) and give both models some time (say 1 hour) to exit the water, using identical gases and switch depths, or with rebreathers, identical diluents and setpoints. Then play with the models until each gave its best shot at a 1 hour runtime. Then let several thousand people dive each, and collect the data. Somehow I don't think this will ever happen - so instead we're going to keep getting threads like this one.

To Lance - I'm not sure what AM's point is since EVERY model has been tweaked following DCS, Haldane in particular (of which Buhlmann is a subset, sort of Haldane D or E or F - God knows how many revisions it's been through following ongoing evidence of excessive DCS risk). And good luck replacing VPM with PURE Buhlmann (which fortunately, neither you nor AM are dumb enough to do).

__________________
Andrew Ainslie

UWSojourner

I suppose, AnnMarie, if you're requiring "test methodologies" and "full data" for asserting anything about an algorithm, it's fair to ask you for the same. So please, provide us with your "test methodologies" and "full data". Remember, anecdotes don't count as "full data".

Can you please provide any hard evidence that misreporting of data sources and methodologies has occurred. How? Who? When?

Also, please describe the "very foundations" of VPM that have been modified based on DCI reports.

Hanssing

WEll seeing that the underlying mathematical model is the infact a buhlman, and no matter what the stops generated are, the tissue tracking mechanisms are working the same...
Only the generated stops are affected by the GF-settings, so your above anecdotal stament only shows that you used a vastly more liberal compartment-model (M-value matrix) compared to what-ever the VR3 was using, or more liberal GF-values.

No matter how you set your first GF the deco will allways be longer compared to both factors being eaquel (say 100/100) ie. pure buhlman, since the faster compartments will not be pushing the gradients and the slower compartments continue to ongas. If using GF setting wich introduces deepstops causes people to get bent compared to non-deepstop setting (GF-factors being equel), this only indicates that the underlying compartment-tracking and M-values does not model our bodies "correctly"....

Regards
Nicolai Hanssing

__________________
Woohooo - I can change my rEvo!
Its going to be bitchin' tricked out piece of gear