Deep Stops Increases DCS

[rossh]

The heat map below shows the supersaturation patterns for......

Wrong Kevin. You have to stop posting these blatantly false claims. You are NOT showing supersaturation.

.....I have to agree. "Green" is not supersaturation....

... The heat map is not predicting risk. Ah, yes. Of course. It never claimed to predict risk.

So Kevin, you see how confusing and invalid your graphs are? If 50 to 75% is "not" supersaturation, then in means that half of your graph details is invalid.

For a science chart, with a 50-75% error - not good enough.

.
A color isn't supersaturation. The color represents a certain relative level of supersaturation between the profiles shown. in this case the lighter green is 50% of the high. It's not a range. The colors start blending at anything lower or higher.

So in summary,

Your heat graphs do not show any meaningful supersaturation information.

Your heat graphs do not show any meaningful presentation of risk.

They are on average, 50 % wrong then.

They are being compared to a complete unknown value.

They are highly misleading, with no actual data points,

They ignore or violate the actual formula for supersaturation.

And you use them to compare unrelated objects, with no actual meaning of supersaturation or risk in the result.

......

Finally - some truth.

.

 

[leadduck]

A1 and A2 hit ceiling under all settings of VPM. Just checked with multideco. VPM asks for stops before first stops of both A1 and A2 profiles. So if I understood your tesis correct it rejects both. How does Subsurface implement VPM do not know. But know original fortran VPM-B program rejects A1 and A2 sure with properly implemented multilevel - original program does not check for hitting ceiling within multilevel.

Thanks Igor. I took the time today to get Erik Baker's VPM-B Fortran code run on Linux. Now I see what you mean, I can't see whether NEDU A2 hits a ceiling in VPM-B when I enter it, because the program doesn't insert stops into multilevel profiles. I can only see that the first stop of a minimum-runtime profile for a 30min@170fsw dive would be at 100fsw, but that's irrelevant to my analysis.
So I guess that UWSojourner's approach of comparing supersaturation levels is the better way to compare profiles and models.

Read the corelation paper I posted and I think you will find answer. Not direct but can be extracted from material.

I read the paper. I think the analysis hinges on B.Wienke's definition of the risk function ρ(κ,t). He only shows the correlation of models with his particular choice of risk function, but cannot show which model best predicts actual DCS risk in experiments. His results only show that he picked a risk function that is similar to RGBM's predictions.
The authors should have made it more clear in the conclusion, that this analysis Table 4 does not mean that RGBM or VPM predicts DCS risk any better or worse than USN or ZHL16.
His choice of risk function is a bit strange by the way. It seems not to consider the fact that different compartments can endure different levels of supersaturation, in particular the fast ones are less sensitive than the slow ones. Correlating with this risk function favors models that set wrong priorities between fast and slow compartments.

 

[leadduck]

The chart below shows the integral supersaturation by compartment for the 4 profiles.
View attachment 378448

I like this style plot; the axes are clearly labeled with numbers.

Some more ideas:

• VPM-B+7 may be valid but is not used in practice. VPM-B+3 has a shorter run time; why not pad it with surface time? Decompressing at 0ft is not different from decompressing at 10ft. You can compare any profile, and integrate supersaturation until the maximum dive time.
• I may have missed the exact definition formula of Integral Supersaturation in this lengthy thread, i.e. the number plotted here, where can I find it? (it's roughly clear, but is there some exp in the integrand like in B.Wienke's risk function?)

ZHL16 is often used wrongly today with computer software and needs a bit caution. ZHL16 was meant to be used to create tables. If you look into Bühlmann's tables of 1986 ( http://www.seveke.de/tauchen/download/tabellen/html/vdst/tabellen.html ) you'll find them to be much more conservative than what you get with ZHL16 and GF100/100. It's more like GF85/85, and the recent tables (Deco2000) are more like GF80/80. Plus, when reading these tables for dives with cold and exertion, the rule is to add 50% to the bottom time to add conservatism.

I don't think Bühlmann ever suggested using ZHL16 GF100/100 for dive table creation or for dive computer use. Saying that nobody dives that anymore today, and thereby implying that this was done in the past, just shows a deep misunderstanding of ZHL history.

So when getting a deco profile to the NEDU case 30min@170fsw working dive with temperature stress, I suggest using GF80/80 with 45min@170fsw. That's hopefully close to what Bühlmann would have done in 1990; it gives a total runtime of about 200min.

 

[UWSojourner]

So in summary,

Your heat graphs do not show any meaningful supersaturation information.

Your heat graphs do not show any meaningful presentation of risk.

They are on average, 50 % wrong then.

They are being compared to a complete unknown value.

They are highly misleading, with no actual data points,

They ignore or violate the actual formula for supersaturation.

And you use them to compare unrelated objects, with no actual meaning of supersaturation or risk in the result.

......

Finally - some truth.

.

Whatever helps you make sense of whatever world you've trapped yourself in.

The heat maps visually compare the supersaturation patterns of the profiles they show. They shed some light on the applicability of the NEDU study, at least for those who've not been VPM-Blinded.

They have limited utility outside this specific exercise and only have value because we have two HARD DATA POINTS in the NEDU study -- the deep stop profile with 5% risk and the shallow profile with 1.6% risk. Without that study we would not know that the pattern of supersaturation shown in the A2 heat map was inferior to that of the A1 profile. In reality, they only present in another form information already published by Doolette, et al, in the original study. But for me they helped explain what that study was trying to tell us.

 

[rossh]

So I guess that UWSojourner's approach of comparing supersaturation levels is the better way to compare profiles and models.

Please... use the correct terms.. UWSojourner's heat graphs do NOT show supersaturation. They show tissue gas, which... is mostly useless information, and actual decompression stress and supersaturation cannot be seen in tissue gas graphs.


The most important part of a plan is the actual supersaturation, because that is where the extra-vascular tissue bubbles that are thought to make DCS, and will form and grow.



REAL supersaturation information:

 

[UWSojourner]

I like this style plot; the axes are clearly labeled with numbers.

Some more ideas:

• VPM-B+7 may be valid but is not used in practice. VPM-B+3 has a shorter run time; why not pad it with surface time? Decompressing at 0ft is not different from decompressing at 10ft. You can compare any profile, and integrate supersaturation until the maximum dive time.

This would be a concern. Integral supersaturation is a useful "index of decompression stress" when comparing dives that vary only in the distribution of stop time. Doolette has stated this many times (please see this post). It could, and I'd think often would, mislead if used to compare vastly different profiles. That is part, actually a lot, of my reluctance to simply publish the heat map program since in its current form I have no way to limit inputs.

I may have missed the exact definition formula of Integral Supersaturation in this lengthy thread...

Integral supersaturation is simple . If you are exposed to 1000mb of supersaturation for 5 minutes, then that would be 5000mb-min of supersaturation exposure (or integral supersaturation). It's just supersaturation x time. It's usually expressed in the form of calculus (integration from 0 to infinity of supersaturation at time t) so can look difficult, but its just a measure of your exposure time to supersaturation.

You'll find integral supersaturation used as part of larger algorithms, whether in Wienke's risk function, VPM's critical volume algorithm, US Navy probabilistic decompression models, etc.

 

[UWSojourner]

The issues relating to whether VPM-B+7 "protects the fast tissues" but the NEDU deep stop profile did not were thoroughly vetted in the deep stops thread.

See these 2 posts.

Post 1
Post 2

 

[boulderjohn]

---

---

A ScubaBoard Staff Message...

We are doing our best to keep this very important discussion at a high level. We have had to remove a couple snide/snarky comments that were not at all helpful today. Please continue to contribute substantive posts that contribute to participant understanding, but please refrain from comments that are not helping us maintain the appropriate tone.

---

 

[rossh]

The issues relating to whether VPM-B+7 "protects the fast tissues" but the NEDU deep stop profile did not were thoroughly vetted in the deep stops thread.

WARNING.

Ignore all references to VPM-B +7. It is FAKE !

They could not make any real connection or faults to VPM-B, so they just made up a FAKE profile.

Pathetic! A deliberate deception. Do not let them trick you.

.

 

[UWSojourner]

WARNING.

Ignore all references to VPM-B +7. It is FAKE !

They could not make any real connection or faults to VPM-B, so they just made up a FAKE profile.

Pathetic! A deliberate deception. Do not let them trick you.

.

Your objection was already thoroughly vetted. See this post. But the relevant chart showing the conservatism level VPM-B+7 is below. As was thoroughly hashed out a couple of years ago, it is an odd argument to say that VPM-B works at lower conservatism levels, but not higher conservatism!???

The real issue is that, regardless of conservatism level, the results of the NEDU's study would lead us to the conclusion that VPM-B skews too much decompression time toward deeper stops.