Friday, May 27, 2016

Do male rats using cell phones get healthy tumors?

NBC News and others just informed about partial results of some new $25 million U.S. government of health impact of cell phones. The full results should be out in 2017.

They claim that rats get brain and heart cancer out of the cell phone radiation. But only the male rats. And those tumors are great because they increase the rats' life expectancy! ;-)

Needless to say, I believe that the tissues view this radiation basically as a thermal one, very slow, not enough to ionize something, and the cells would have to deliberately try to detect waves at this frequency and start cancer if they detect something. Why would they be doing it?

These bizarre statements made me quickly look at the PDF file. What sort of an effect did they see?

They tried some radiation exposure on groups of 90 rats, male and female ones, and so on. The exposure was 0,1,2,3 units, you may find the details there. On page 58, I found a table showing the numbers of heart tumors.

For the 8 groups of 90 rats, male and female, with increasing radiation, the number of heart tumors was\[

0,2,3,6; 0,2,0,2

\] So there's no clear correlation for the female ones. What about the male ones? There seems to be an apparent increasing function. The authors claim that the \(p\)-values are as low as 0.01 and they appear several times.

As far as I can see, each group of 90 mice may be modeled by a random group where the tumors strike via Poisson distribution with the average number equal to 2 or so. The Poisson probability that you will get \(k\) – which should be the numbers 0,2,3,6;0,2,0,2 – is\[

\frac{\exp(-2) 2^{k}}{k!}

\] In a self-explanatory way, you may list the probabilities via the Mathematica command
Table[{k, Exp[-2.]*2^(k)/k!}, {k, 0, 7}]

{{0, 0.135335}, {1, 0.270671}, {2, 0.270671}, {3, 0.180447}, {4, 0.0902235}, {5, 0.0360894}, {6, 0.0120298}, {7, 0.00343709}}
and the result was copied-and-pasted. Great. So the appearance of 0,2,3 is mundane, the appearance of 6 comes with the probability 0.012, about 0.017 that it's "at least six".

But the paper had about 10 places where such a 6 could have occurred, so the probability that such an elevated figure appears somewhere is something like 0.17, so the "discovery" is basically a 1-sigma effect once the look-elsewhere effect is incorporated. It's just plain ludicrous to take it seriously. And I am not even officially accusing them that they tried several groups of 90 rats before they finally got a "six".

If the radiation hurts, shouldn't the response be increasing (approximately linearly) with the exposure? I have nothing against the rats but can't they simply pump a 20 times greater amount of the radiation to see whether most of the rats develop tumors?

You know what would happen, right? Nothing would change. Theories of this kind could be easily demonstrated experimentally if they were right. But they are not right. So people who make living out of this fearmongering ($25 million is not bad) prefer lower doses and cherry-picking plus overhyping of one-sigma "signals".

It's shameful that this kind of garbage is funded by the governments or pretended to be scientific research.

No comments:

Post a Comment