Bubble model vs. Gradient Factors redux

[ChuckP]

Without reopening old, old discussions before the data arrives, isn't the same run time the point? There have been too many arguments about the most "efficient" or the most "conservative" profile. But with two dives of the same run time, "Which one is at greater risk?" is a fair question.

I think no - diver safety should be the point.

I think that's where some of the arguments have been centered on, getting the same run time close has lead to some saying the test was skewed....

There are a bunch of dives made using GF's - I think a lot of 50-100m dives are made using 40-50 GFlo and 70-80 GF hi. We dive our GF's to give us a margin of error, a safety factor if you will.

I assume there is some factor that can be applied to a deep stop model that will allow a similar margin or error/safety factor - I understand bubble models fairly well but have zero experience in diving or planning them. There has to be guys out there diving XYZ to these depths.

I'd like to see the run time ignored, let's not worry about efficiency or conservativism - let's not hurt people but let's test what people dive, not some theoretical thing that no one does. You deep stop guys, bring something that you feel is 75-80% conservative, the GF guys decide on a GFLo that is being used and set the high at the 75-80 number. Plan the dives with the same gases (i understand with deeper stops, the initial deco gas ideal choice may differ but you have to work this somehow), do the dives in a chamber and measure with doppler however you can during the dive - all done in a chamber of course. Let the run times be what they are, it's what we dive - I have to believe they'd end up +/- 15% anyways.

Point being - test plans that are actually dived, compare the induced stress that each plan develops during and after the dive. Tech dives are spread thru such a broad range and there's so many factors that are really understood that effect the induced stress, it would be nice to have a side by side comparison of two different methods showing bottom to boat to beach so that we could make better decisions for diver safety across a broad range of conditions.

I confess, I haven't watch the video up top yet, I read thru the other chat group and a few of the studies - I hope to have watched that by this evening.

I wonder if we would see that one method is better than the other when applied in different conditions. In Mark's book, he refers to a deep type stop study that showed better results, depth was 25m and these were NDL dives - Simon also refers to two other studies that refute that also, I have to read those too... Information, we need more information but the information has to be valid for the way we dive......

 

[Storker]

diver safety should be the point.

Well, you have to choose a criterion on which to compare algorithms. To me, it seems rather sensible to choose "same run time" as that criterion and compare f(DCS). Alternatively, if you aren't allowed to bend a significant number of your research subjects, f(bubble score >2) because f(bubble score >x) correlates quite nicely with P(DCS).

 

[dmaziuk]

I think no - diver safety should be the point.

Don't dive and you'll be safe.

If you do dive, you'll have certain constraints. Like, maximize bottom time to do useful work and minimize ascent time to get your boat out of the harm's way ASAP (that what Navy usually does). You can't run meaningful comparisons based on "diver safety", you can only compare "bubble grades" after the dive.

 

[ChuckP]

Well, you have to choose a criterion on which to compare algorithms. To me, it seems rather sensible to choose "same run time" as that criterion and compare f(DCS). Alternatively, if you aren't allowed to bend a significant number of your research subjects, f(bubble score >2) because f(bubble score >x) correlates quite nicely with P(DCS).

Isnt one of the selling factors in bubble models is that it gets you out quicker?

I just disagree, I’d like to see models compared under real dive plans - show me the stress each one induced where So that I can make a decision based on the dives I do. One model may hang 10 minutes longer but show more stress at initial stops.....

Maybe you have to bend people to really know, not sure, I’d hope that Doppler results can show stress before bending.

 

[rsingler]

I’d like to see models compared under real dive plans - show me the stress each one induced where So that I can make a decision based on the dives I do.

I really think that's what this upcoming study tried to do. I believe the bottom times were the same, and were realistic.
The run times were close to identical, for purposes of comparison. And what was chosen was a pair of gradient factors that, for the same deco time, differed only in where the distribution of deco lay. One was deeper; one was shallower. That's why the GF Hi's were different too: in order to have the same total deco despite deeper initial stops, a higher GF Hi was required. You saw the difference.

Now @scubadada has a point - we've just changed TWO variables. For the next study, how about locking GF Hi, accept the longer run time for the deep stops dive, and with the same GF Hi, see how they compare? A valid question.

 

[scubadada]

Hi @rsingler

As was discussed in the video, there's not much money to support decompression research these days, not the Navy, not the oil industry. I'm glad to have whatever new data we can get.

Being constrained by run time controls that factor, but limits your ability to adjust deep stops, GF low, and surfacing GF, GF high. Controlling GF low or GF high makes it impossible to control run time.

Take the NEDU profile as example, 170 ft for 30 min on air. Suppose you want to compare a change in GF low from 20 to 50 or a change in GF high from 70 to 85. I ran these on MultiDeco with a stop time of 1 min:

Comparing group 1 to group 2 or group 3 to group 4 would tell you what the effect of changing the GF low was. I'm not smart enough to know if they would give exactly the same result. Comparing group 1 with group 3 or group 2 with group 4 would tell you the effect of changing the GF high was, same caveat. I believe that if you require that the run time be the same, you will have to adjust both GF low and GF high and make the analysis more complex. At least in the example I chose, the change in GF low changed the deco time much less time than the change in GF high, I had not noticed that.

I think that you and I are in the same chapter if not on the same page, we would just like to know how this works

 

[dmaziuk]

For the next study, how about locking GF Hi, accept the longer run time for the deep stops dive, and with the same GF Hi, see how they compare? A valid question.

Isn't this where Spisni study went, kind of? They had longer ascent times for ratio deco profiles, or am I mis-remembering?

 

[ChuckP]

Don't dive and you'll be safe.

If you do dive, you'll have certain constraints. Like, maximize bottom time to do useful work and minimize ascent time to get your boat out of the harm's way ASAP (that what Navy usually does). You can't run meaningful comparisons based on "diver safety", you can only compare "bubble grades" after the dive.

Aren’t bubble grades or scores a way to evaluate diver safety - dive safety?

Does a 50/90 dive produce a worse bubble score than a 50/70?

So how does a deep stop plan compare to a 50/75 plan? Compare them as dived, as the planner planned them - no adding or decreasing run time.

Early on in table development, they noted that what worked at one depth range didn’t work so well at another depth range - they altered the tables. They didn’t know what they didn’t know.

Im certainly not arguing one side or the other, it’s just I think there’s some things that we don’t know and as I venture from the 75m depth to 100m, I’d like more real world knowledge, not hypothetical extrapolations - which this DAN study might have done, I just haven’t had time to watch it

 

[Storker]

Isnt one of the selling factors in bubble models is that it gets you out quicker?

Well, if you want to compare models, you can only change one parameter at the time. Otherwise, you can't know where the difference is. That's research 101.

So, if I were to design an experiment, I'd keep the bottom time and runtime constant and only change the ascent schedule. So I'd choose a VPN conservatism level and a GF combo which got the subjects out of the water at the same time. And I'd choose a GF combo which gave a significant difference in deep/shallow stop distribution compared to the VPN schedule. Otherwise, the difference in schedule would be too small to give any meaningful difference in the results.

And then add the ethical issues of deliberately bending a bunch of divers (which, generally, only NEDU is allowed to, and even they, only to a very limited degree), the statistical problems if your cohorts are too small and the issue of financing the study... Well, damn, I'm pretty happy I'm not a hyperbaric scientist trying to find answers to those questions.

 

[dmaziuk]

Aren’t bubble grades or scores a way to evaluate diver safety - dive safety?

It's a stretch. They are a proxy measure of decompression stress post-dive. Translating that to "diver safety" would need for starters a formal definition of "diver safety". Then you'll need to correlate that to Spenser scale. And so on.

(PS and note that "decompression stress" isn't defined all that rigorously either.)