Decompression Modeling

[ScubaFeenD]

That said, I seem to make it out to LA once every year or so these days, so perhaps we can work on a) ;-).

Are you inviting me to join you on a dive trip to socal. As it just so happens my schedule is just about to open up.

 

[ScubaFeenD]

I believe to some degree, the distribution is known -- I think there are definitely depth/time combinations where you can say pretty much for sure that if you do them, you will be symptomatic. And there are some where, within 99% probability, you won't be. The problem is that we are trying to define in fine a process which is only understood probabilistically, and that type of understanding is poor at defining probabilities for any individual.

It's like coronary risk factors. They're great at predicting cardiac risk likelihood in a population, and all but worthless for assessing it in individuals.

Your right regarding populations and individuals. However, there seems to be fairly decent disconnect between the models of decompression physiology and the individual hits. Sure, from what we know is going on and think is going on with the physiology we are able to predict with some certainty what will definitely give a DCS hit (subclinical or full-blown). However, there still seems to be really poor grasp of the real pathology of the disease--unlike heart disease. Sadly, while it is an extremely interesting subject to understand, it is a low priority field for research.

 

[gsk3]

I believe to some degree, the distribution is known -- I think there are definitely depth/time combinations where you can say pretty much for sure that if you do them, you will be symptomatic. And there are some where, within 99% probability, you won't be. The problem is that we are trying to define in fine a process which is only understood probabilistically, and that type of understanding is poor at defining probabilities for any individual.

Maybe, depending on how far we're going with "to some degree." If the latest in deco models can't produce a profile that a DIR diver is happy with (assuming the DIR method is closer to the "truth"), no matter what numbers you tweak it with, then it's not a matter of parameterization but of the true model being known. I'm kind of muddying the waters here between models and distributions; you could argue that for a given deco model, we have a good sense of the distribution of failures outside that model (in the extreme case, just assume it's a logit or similar binary model for failure vs. not) even if the model itself is wrong. That might be true, and the benevolent monopoly on diving medicine that is the DAN hotline means that we probably have a good sense of the numerator, but we still have no idea about the denominator--how many technical dives are done in a year globally?

I guess I am made skeptical by quotes like this that I ran across right before my earlier post:
"Unfortunately, the recommendations for decompression are largely empirical and not always reliable. This is confirmed by the finding that more than half of the DCI cases managed by DAN worldwide over the past several years have not been associated with an obvious violation of decompression procedures, dive table, or dive computer limits; they have been 'unpredictable.'" p18, proceedings of the Future of Diving conference, 2009. Lang and Brubakk ed.

It's like coronary risk factors. They're great at predicting cardiac risk likelihood in a population, and all but worthless for assessing it in individuals.

Exactly. What's interesting here is that the individuals have looked at what the models produced, seen how much variation there is, and found a factor that they think explains some of that variability (the whole sub-clinical DCS, individual adjustments to deco curves bit). In theory, it's even testable and therefore could be incorporated into later models. In practice, it would take a very large study and divers who are experienced enough to have collected data on their own feelings of fatigue vs. profiles.

 

[gsk3]

Was merely saying that I have family in LA and find myself there frequently. If you happen to find yourself on the same plane next we go out, it won't be my fault that the family vacation got crowded out by diving, will it?

there still seems to be really poor grasp of the real pathology of the disease--unlike heart disease.

C'mon. We know all we need to know. It's about bubbles. And complement. And stuff.

 

[rjack321]

Perhaps it really does come down to -- "If you do the crime (deco) you must do the time" and how you do the time isn't as important as just doing it somewhere.

Experience has shown this to be roughly true for going 5-10mins over MDL or modest 1 or possibly 2 deco gas dives. But doing 50% too much time on 35/25 and 50% too little time on O2 is a recipe for disaster. The deco time penalty is only swappable within the profile to a degree. You can illustrate that for yourself by creating multilevel profiles in V-planner and seeing how the shallow stop time decreases as you add time at O2 window or intermediate stops.