Deep Stops Increases DCS

[rossh]

I'll let diver's decide for themselves. For those interested the presentation here is a good elaboration of the issues.

It's you quoting him, and him quoting you - a circular argument. Still the same invalid comparison methods used.

Completely missing from his tape, is the concept of tissue gradient vs supersaturation. He joins them together, as a simplicity. But it's a mistake, as it invalidates most of the rest of his argument. He tries to make the same non-existent connection to tech tech profiles that you have done. Plus we can't overlook the deliberate mistakes he makes in VPM-B history, design, or VGE origins, and impact, etc.

.

 

[The Chairman]

Honestly, you guys are talking so far above me, that for once I feel rather flummoxed. Can I ask how the "heat graph" was created? What are we actually looking at?

I don't mean to demean... but the various decompression theories seem to be nothing but SWAGs: a Scientific Wild Ass Guess. You're not actually measuring the various tissue compartments, you're only guessing how much N2 has been dissolved into them. I just want an algorithm that works consistently with little chance of getting bent. I want that to be in a little black box on my wrist that I can consistently depend on.

 

[leadduck]

rossh, isn't the cause of all of these endless threads here and elsewhere that you sell a product V-Planner that relies on VPM? Now it turned out that much of the underlying technology that you founded your business on is poor. Well, that's really inconvenient, but **** happens; get over it. Of course you can continue selling VPM planning software. I don't expect vendors to stop selling RGBM or VPM dive computers either, they'll sell it as long as there's a market for it.

In the future however you have to expect a decreasing number of users of VPM planning software. Maybe we'll see also a broader choice of methods in planning software, such as linear-exponential VVAL18, or better ideas than GF for tweaking Buhlmann.

 

[Storker]

Can I ask how the "heat graph" was created?

I see them as 3D graphs. Deco time on the X axis, compartment # on the Y axis and calculated supersaturation coded by the color. You can plot the same data as contour maps, or as 3D bar graphs, but color-coding is a very easy and intuitive way of representing the magnitude of the Z data in a 3D graph. I prefer to have a legend showing which Z values that corresponds with which colors, but for a qualitative representation, that isn't necessary.

Here's a similar type of plot:

 

[rossh]

rossh, isn't the cause of all of these endless threads here and elsewhere that you sell a product V-Planner that relies on VPM? Now it turned out that much of the underlying technology that you founded your business on is poor. Well, that's really inconvenient, but **** happens; get over it. Of course you can continue selling VPM planning software. I don't expect vendors to stop selling RGBM or VPM dive computers either, they'll sell it as long as there's a market for it.

In the future however you have to expect a decreasing number of users of VPM planning software. Maybe we'll see also a broader choice of methods in planning software, such as linear-exponential VVAL18, or better ideas than GF for tweaking Buhlmann.

We sell more than VPM. MultiDeco is our main product - with two models and room for more. Happy to add any new and validated info.

(added..) I have actually built a better ZHL model, just to see what it does. It has integral and properly handled gradient changes, and separates N2 and He adjustments. Its scales properly without the exponential error of the existing GF. But why bother?

VPM-B is still popular and works perfectly well. No faults at all with it. In fact the latest "new" gas kenetics, have almost now difference to it.


I'm going to toot my horn here - in recent years, the conversation has been about this supersaturation processes - our program is the only one to give divers the tools to do this, as shown above. Then the conversation was about "new" kinetic" helium rates. Our V-Planner has that adjustment in built - make you own settings and experiment all you want.

Then just this year, the thoughts have shifted again towards Inner ear DCS and the possible related excess accelerated decompression issues. Again our programs have had this regulating ability for 12 years now. So we make the tools that let divers explore all this new and old theory, as best we can.

I attend the important meetings and gather info, I'm a member of various societies, and keep an eye out for new and interesting stuff to be added.

VPM might not be your favorite model, but its still the most current and correct model we have in use today. No added fudges, makes perfectly good deco plans straight out of the box, that anyone can use. In contrast, real ZHL was dropped a long time ago, and most plan types today, have adopted the bubble model attributes - in larger or smaller amounts. There is the military plans, but they tend to be very shallow, and not something that tech divers want to do. We are also limited by the need for trimix planning, which most military do not have. So whats next? The hybrid model that seems to be in favor, cannot be duplicated in pure dissolved gas theory form. So it needs a physical component. Fiddling the half time is not going to get there either.



You ask why does this argument keep going? See the two long contrasting posts above. I'm defending the VPM model from being miss - characterized, or described in unfair ways. Some others want to shift deco towards a different place. Some want to unfairly criticize VPM for things it has not done, and some will apply as much voodoo scare tactics as possible to frighten people into moving.


But consider this - the dive spectrum has almost been fully explored already. The early models were short and shallow, then shallow and longer. Many dissolved (theoretical) models occupy the middle ground. Then Pyle stops, bubble models went to a middle to deeper area based on microbubble physics. Then some ad-hoc methods went even deeper, almost to the bottom So what's left to explore in terms of pure decompression design? Not much left. Today's new direction - ever longer in the shallow sections, is not exploring - it's just padding.

Just about any plan we use today is fine, because they all have more than enough deco for the conditions. Looking at DCS numbers, they have been steady / low for 15 years now. And that corresponds to three things - much greater discipline in procedures, thanks to the DIR movement, and 2/ the deep/er stop profiles, with modest to slow shallow sections (typical of all models / plans today), and 3/ much better avoidance of known bad deco practices.

And were are we heading? Away from pure decompression models, and towards extra safe added on padded margins and more (unnecessary) deco time, for just in case purposes. Are they any safer? The numbers do not bear out any real changes. Are they actually needed? I don't think so, but anyone is welcome to pad up the deco all they want or need or desire.

.

 

[leadduck]

I don't mean to demean... but the various decompression theories seem to be nothing but SWAGs: a Scientific Wild Ass Guess. You're not actually measuring the various tissue compartments, you're only guessing how much N2 has been dissolved into them. I just want an algorithm that works consistently with little chance of getting bent. I want that to be in a little black box on my wrist that I can consistently depend on.

Well that actually summarizes it pretty well. All decompression models use a primitive model for very complex physical effects in the human body.

Simulating a complex system (such as the human body) with a simple model (such as Buhlmann's or VPM's independent compartments) is fine, as long as one very important procedure is part of it: calibrating the model parameters and validating the simulation error with controlled experiments. That is the most important step of modeling.

Buhlmann's model is no rocket science; the value of it lies in the fact that M-values were determined by experiments. Same for Navy tables, DSAT RDP, and many others. It's all about calibrating a simple model by experiments.

I found it interesting to read Dr. Yount's 1986 paper "On the use of a bubble formation model to calculate diving tables" under this aspect. His motivation was: reduce the number of model parameters that need to be calibrated by including more physical effects in the model, in order to reduce calibration effort and make the model valid over a larger range of variables such as depth and duration.
That's quite ambitious and I wouldn't say he succeeded. By introducing a bubble model for every compartment, the number of parameters goes up, not down. To solve that problem, he decided to use the same values of bubble model parameters for every compartment. That's pretty rough and I think this is what killed the potential advantage of his approach; as a side effect it produces deep stops.
Secondly, the reduced parameter set didn't get calibrated by experiments. In the paper, the bubble model parameters are chosen so that the total decompression time matches the US Navy table. The shape is different however: longer deeper stops, shorter shallow ones.
Seeing the results of this paper, the conclusion should have been that this model has a significant error, probably caused by an oversimplified parameter structure and lack of calibration by experiments. Instead, some misinterpreted artifacts of the model error as a new discovery (shifting deco time to deeper stops is good), and used it for diving without calibration.

 

[rossh]

Honestly, you guys are talking so far above me, that for once I feel rather flummoxed. Can I ask how the "heat graph" was created? What are we actually looking at?

I don't mean to demean... but the various decompression theories seem to be nothing but SWAGs: a Scientific Wild Ass Guess. You're not actually measuring the various tissue compartments, you're only guessing how much N2 has been dissolved into them. I just want an algorithm that works consistently with little chance of getting bent. I want that to be in a little black box on my wrist that I can consistently depend on.

Hi Pete,

No model tries to make a true representation of the actual physiology processes. Imagine if such a tool was possible - real time circulation monitoring - then medicine would adopt it for a thousand more important uses, than just divers fiddling with profiles.

And further imagine what is going to happen when we get a real time probe stuck to our skin, and transmits to a updating display computer. Deco times will probably get much, much shorter..... any takers?


In decompression, there are big pressure changes, so the model does not need to be intrinsically accurate - just accurate enough for getting too the wide grey area that's sufficient to do the job. Models only try to monitor the onset of harmful conditions, but nothing else. Getting it close enough is pretty easy done today, and the only difficulty seems to be delivering the desired preference for style of successful deco (deep / shallow / long / skinny, etc. [sounds like coffee]).

( I see leadduck has made a nice post on this above).


What is a heat graph? This will make steam come out of Kevin

You have seen many tissue pressure graphs in the past - 16 parallel lines going up, then down. That's the raw tissue pressure values shown there (on / off gassing). Tissue pressures going up / down, are 90 % noise - no actual decompression limit is reached for most parts within for most of the time.

Heat graphs is that same info, segmented into each stop level, during the ascent. Then the surface part shows off gassing (not supersaturation). Then it draws on /off gassing which gives the bias in colors. But no actual decompression limit data is shown.

He then goes further and compares dive A and B: which could be two lame dives, or two serious dives, or one of each. The point being that it shows the on / off gassing information only, and not supersaturation or decompression limit information. Its all very interesting, pretty, but useless information on its own, because decompression limits are built from a larger set of information and pressures.


To get actual decompression limiting data values, it needs to be computed and extracted. We need to know exactly where the dive fits in the aggressiveness scale, so we need to look actual model calculated limiting pressure supersaturation values, and that is what is shown in my programs graphs. Its also how the deco model limits its ascents. It also where GF is based too.


... (awaiting the rebuttal.)

 

[rossh]

...... Seeing the results of this paper, the conclusion should have been that this model has a significant error, probably caused by an oversimplified parameter structure and lack of calibration by experiments.....

Yes, that is the reason for VPM-B. To restore a missing parameter in the central equation.

 

[RyanT]

Ross has been bashing heat maps as meaningless and unscientific. This is simply not true. They have real value in allowing us to visualize complex patterns. We use them all the time to look at patterns of genetic similarities among individual and complex patterns of gene expression in organisms. Like any type of mathematical modeling, it is important to understand how the heat maps are generated, but they are certainly not uninformative or "unscientific."

 

[rossh]

Ross has been bashing heat maps as meaningless and unscientific. This is simply not true. They have real value in allowing us to visualize complex patterns. We use them all the time to look at patterns of genetic similarities among individual and complex patterns of gene expression in organisms. Like any type of mathematical modeling, it is important to understand how the heat maps are generated, but they are certainly not uninformative or "unscientific."

Well I'm not suggesting they have no use elsewhere.. . They just have very limited use for a dive decompression, and they show no valuable or controlling limiting data.

.