Dive computers are untested and unsafe?

[Jax]

Wait - didn't the Navy use trial and error to come up with an algorithm?

And then the Navy made their tables from the algorithm?

And the agencies looked at the Navy tables and SWAGged their own to that?

And along comes computer makers that massage those algorithms according to a perceived value of conservatism?

And this massaged algorithm is applied to a SWAGged population that is someone's best guess of average?

All to be tested against somebody's specific history, genealogy, physiology, and given levels of fitness, alertness and general condition at some moment of time.


Yeah. What DSix said. It's all VooDoo.

 

[boulderjohn]

Several of the tables have been extensively tested. I don't know what testing has been done on any of the computer algorithms. Rubicon might be a good source for information about that, if you can figure out how to phrase the search.

There are just two tables/models based on experimental data? What about Haldane, Hawkins, Yarbrough, Van der Aue, Buhlmann, BSAC, DCIEM, NAUI, and VPM?

A while ago I embarked on several explorations of this topic, conferring regularly with Gene Hobbes of Rubicon as I did so. He gave me a lot of help.

The truth is that a lot of these have not been tested to the degree one might think they have been. Bühlmann more than most. Of the others, some have not given any real indication of where they came from. Here is one example. On one of my investigations, I looked at DCIEM for reasons that are not important now, and I learned that very little is known about the theory behind that or the indications of its effectiveness. One of the people who developed it had published some papers prior to that work which gives people reason to assume some of that thinking might have gone into it, but no one knows for sure.

As for VPM, RGBM, and other models, there is some research but nothing at all like the work that was done for the PADI RDP. According to what Gene told me, he was not seeing any results much different from the Navy research. There was supposed to be some work coming out in the relatively near future, but to my knowledge that hasn't happened.

We can all go on with our own experiences, but that is a very limited study. For example, the small group of people with whom I have been diving at a small lake in New Mexico has had 6 DCS cases in the last couple of years, a pretty high ratio, but those numbers aren't going into any statistical data base that I know of, so not many people know about them, and they could be a real anomaly that is totally out of character with experiences world wide.

 

[ianr33]

I've been experimentally testing my computers over the last 20 years and several hundred dives.

Haven't been bent yet. Good enough for me.

 

[Dr Deco]

Hello Readers:

Using Theory to Predict Nitrogen Loading

Deco models do not have hard proof behind them. They use the system from John Scot Haldane [1908] with half-time compartments and it is wholly empirical. These “compartments” are not necessarily whole tissues but can be pieces of tissues. Actually, it is not the gas loads that need predicting but rather the tissue micronuclei number. Not accomplishable as of yet.

No one has been able to generate a table from first principals alone [ab initio]. The system works to the extent that they generate tables that work correctly, but there are limits when it comes to extrapolating to deeper depths or longer bottom times. The model also fails when it comes to analyzing why a dive failed. For this latter case, concepts based on bubble physiology work better. [It is this physiological model that I have used here at Ask Dr Deco for the past decade.]

Can One Measure the Actual Nitrogen in the Body?

Currently there are not any means to measure the actual nitrogen loadings in a tissue. What is more important would be a method to measure the bubble micronuclei, but that is another topic. It is true that there is at least one case of a diver who cut a hole in the wet suit so the computer lay next to the skin. The belief was that the computer was actually measuring nitrogen in the body! No, that is not possible.

Do Computers Use Their Own Algorithm?

The majority of computers [when I last checked] used the PADI RDP as the model for the computer; this is because the data can be traced [its pedigree] to actual published data.

You actually can reverse engineer a dive computer and find its constants [limiting gas loadings, NDLs]. You can also reverse engineer the PADI RDP and gets its constants, but you can buy the PADI book on the development and testing of the RDP and everything is written out for you.

In some manner, tables and computers use either laboratory data or data from the field. “Trial and error” is a part of science – but they call it research.

Computers Tables and DCS

The data to date indicate that the incidence of DCS is the same for divers using either tables or computers. A couple of decades ago, I thought that computer users would have more cases of DCS than table users, since they are more likely to be diving to the model limits. However, the conservatism built into the NDLs is very protective. Were there problems, the devices would have been changed.

Dr Deco :doctor:

 

[Thalassamania]

I will trust tested trial and error over science any day.

Hm ... last time I checked "tested trial and error" was science. Has something changed?

Hello Readers:

Using Theory to Predict Nitrogen Loading

Deco models do not have hard proof behind them. They use the system from John Scot Haldane [1908] with half-time compartments and it is wholly empirical. These “compartments” are not necessarily whole tissues but can be pieces of tissues. Actually, it is not the gas loads that need predicting but rather the tissue micronuclei number. Not accomplishable as of yet.

No one has been able to generate a table from first principals alone [ab initio]. The system works to the extent that they generate tables that work correctly, but there are limits when it comes to extrapolating to deeper depths or longer bottom times. The model also fails when it comes to analyzing why a dive failed. For this latter case, concepts based on bubble physiology work better. [It is this physiological model that I have used here at Ask Dr Deco for the past decade.]

Can One Measure the Actual Nitrogen in the Body?

Currently there are not any means to measure the actual nitrogen loadings in a tissue. What is more important would be a method to measure the bubble micronuclei, but that is another topic. It is true that there is at least one case of a diver who cut a hole in the wet suit so the computer lay next to the skin. The belief was that the computer was actually measuring nitrogen in the body! No, that is not possible.

I vaguely remember hearing about a study in the 1980s by Bacharach that involved the saturation of a group of marines at Lawrence Livermore Labs with radioactively tagged nitrogen. I don't remember the outcome(s).

Do Computers Use Their Own Algorithm?

The majority of computers [when I last checked] used the PADI RDP as the model for the computer; this is because the data can be traced [its pedigree] to actual published data.

You actually can reverse engineer a dive computer and find its constants [limiting gas loadings, NDLs]. You can also reverse engineer the PADI RDP and gets its constants, but you can buy the PADI book on the development and testing of the RDP and everything is written out for you.

In some manner, tables and computers use either laboratory data or data from the field. “Trial and error” is a part of science – but they call it research.

Computers Tables and DCS

The data to date indicate that the incidence of DCS is the same for divers using either tables or computers. A couple of decades ago, I thought that computer users would have more cases of DCS than table users, since they are more likely to be diving to the model limits. However, the conservatism built into the NDLs is very protective. Were there problems, the devices would have been changed.

Dr Deco

 

[martinskarin]

Do Computers Use Their Own Algorithm?

The majority of computers [when I last checked] used the PADI RDP as the model for the computer; this is because the data can be traced [its pedigree] to actual published data.

You actually can reverse engineer a dive computer and find its constants [limiting gas loadings, NDLs]. You can also reverse engineer the PADI RDP and gets its constants, but you can buy the PADI book on the development and testing of the RDP and everything is written out for you.

In some manner, tables and computers use either laboratory data or data from the field. “Trial and error” is a part of science – but they call it research.

Last time I was checking my non deco times in my Suunto Zoop (when not having used it for a while) I remember it was corresponding more or less exactly with the RDP so at least for Suuntos computers I can aggree on that.

How the algoritm for when the computer goes into "Deco mode" (if by mistake overstaying or something) I have no clue about, the manual says something about that the computers algoritms is made together with some Dr. something I forgot the name but can check later.

 

[boulderjohn]

Although this is by no means scientific....

Let's look at the overall rate of DCS among recreational divers. Pretty low, huh?

What percentage of those divers over the last decade or so have been done with computers, and what percentage have been done with the tried and true tables?

In all the recreational diving I have done outside of instruction, I don't recall ever seeing anyone plan and execute a dive using the well-tested tables. One day I did see a couple using the Wheel, a sight I thought I would never see. I cannot even begin to estimate the number of divers I have seen using computers.

 

[Jax]

Although this is by no means scientific....

Let's look at the overall rate of DCS among recreational divers. Pretty low, huh?

What percentage of those divers over the last decade or so have been done with computers, and what percentage have been done with the tried and true tables?

In all the recreational diving I have done outside of instruction, I don't recall ever seeing anyone plan and execute a dive using the well-tested tables. One day I did see a couple using the Wheel, a sight I thought I would never see. I cannot even begin to estimate the number of divers I have seen using computers.

Serious question . . . What level of DCS is "low"?

To what are we comparing? Diving 20 years ago? 15 years ago?

Since no one is keeping any real stats (no, not even DAN, because no one knows how many dives are done each year, and that is the best sample size with which to work), and the number of divers is increasing, do we even know what is low?

 

[boulderjohn]

Serious question . . . What level of DCS is "low"?

To what are we comparing? Diving 20 years ago? 15 years ago?

Since no one is keeping any real stats (no, not even DAN, because no one knows how many dives are done each year, and that is the best sample size with which to work), and the number of divers is increasing, do we even know what is low?

Actually, DAN has done a lot of work in this area during the last 15 years or so through its Project Dive Exploration and has a much better grasp on this than you might think. They don't know the total number of dives worldwide, but they know the total number of dives and the total number of incidents of all kind in their representative study. In the latest released edition ([-]1968[/-]2008), it reports on page 6 that the DCS incident rate is 2 cases for every 10,000 dives.

EDIT: I forgot to mention the source. The results are on page 6 of the annual fatality report ([-]1968[/-]2008). You can examine all the charts and graphs there if you would like.

 

[Jax]

Actually, DAN has done a lot of work in this area during the last 15 years or so through its Project Dive Exploration and has a much better grasp on this than you might think. They don't know the total number of dives worldwide, but they know the total number of dives and the total number of incidents of all kind in their representative study. In the latest released edition (1968), it reports on page 6 that the DCS incident rate is 2 cases for every 10,000 dives.

EDIT: I forgot to mention the source. The results are on page 6 of the annual fatality report (1968). You can examine all the charts and graphs there if you would like.

That's certainly the right direction!