From the Washington Post:
Perls and his colleagues analyzed the genes of participants in the New England Centenarian Study, which is the largest study of centenarians and their families in the world. The study involves about 1,600 centenarians and has been ongoing since 1995.
"A lot of people might ask, 'Well, who would want to live to 100?' because they think they have every age-related disease under the sun and are on death's doorstep," Perls said. "But this isn't true. We have noted in previous work that 90 percent of centenarians are disability-free at the average age of 93."
They also noticed that longevity seemed to run in centenarians' families, indicating that genetics must play a role.
So the researchers compared the genes of 1,055 centenarians with 1,267 other people to see whether they could identify any unique patterns. Based on that work, the researchers identified 150 genetic variations that appeared to be associated with longevity that could be used to predict with 77 percent accuracy whether someone would live to be at least 100.
"Seventy-seven percent is a very high accuracy for a genetic model, which means that the traits that we are looking at have a very strong genetic base," said Paola Sebastiani, a professor of biostatistics at the Boston University School of Public Health who helped conduct the study.
I'm not really interested in this topic enough to go find the paper, but my question would be whether the researchers split their sample in half, data-mined one half, then tested their findings on the other half. That's proper research hygiene so that you don't just come up with a lot of small, random associations.
But it's hard to make yourself do it. I remember taking a finance course at UCLA in 1981 where we had to do a SAS analysis of a hypothesis about patterns in the stock market to test the Efficient Markets Hypothesis. So, I typed in from the Baseball Encyclopedia the dates of World Series home games involving the Yankees going back to 1921, and the stock market volume and change in Dow Jones average. The professor had said over and over that you had to divide your sample size in two, but there just weren't enough home games, so I didn't do it and got only a B on the project. (I didn't find any effect on prices, but NYSE volume was down on days of Yankee World Series home games.)
Posts
27 comments:
Bertrand Russell (who lived to 97) said that the key to longevity was choosing one's parent's well.
"Although my own parents both died young, of remoter ancestors I can only discover one who did not live to a great age, and he died of a disease which is now rare, namely, having his head cut off."
Off topic comment: I hear there was some news about immigration by some US official or other. Readers of this particular blog might expect some sor of reaction.
My 95 year old next door neighbor was treated for advanced liver cancer in 1975. She must have something going for her.
Very interesting beginning with an analysis of the service records of WW I veterans found that men who had at least four children before they were 30 were far more likely to live to 100.
"Very interesting beginning with an analysis of the service records of WW I veterans found that men who had at least four children before they were 30 were far more likely to live to 100."
Raising kids is stressful. If you have 4 kids before 30 then you are rid of them by the time you are a relatively youthful 45. If you had no kids before 30 in early 20th century America it probably meant you were undesirable for physical or psychological reasons.
i worked with a girl from Sardinia (longest lifespans in the world) she had a 103 year old great granmother who still went to the market and bought vegetables... she shrugged her shoudlers 'everyone on sardinia has a great grandmother"
According to the NY Times they did have a holdout sample:
http://www.nytimes.com/2010/07/02/science/02longevity.html?_r=1&hp
She found that 150 genetic variants were associated with extreme longevity. She then looked at a different sample of centenarians from those involved in her study and found that more than three-quarters possessed many of the 150 genetic variants she had already identified. The other centenarians had few or none of the protective variants, which means there are many more yet to find, Dr. Sebastiani said.
But an Icelanding "hold-out sample" did not confirm the results:
But Kari Stefansson, a geneticist who has looked for determinants of longevity among the Icelandic population, said of the current study that he was “amazed at how many loci of genome-wide significance have been found in a modest sample size.”
Dr. Stefansson said he had been able to accumulate a larger collection of centenarians, despite Iceland’s small population, because his company, Decode Genetics, has analyzed most of the genomes of living Icelanders and in addition can compute the genomes of Icelanders who lived long ago from the genomes of their descendants. None of the Boston University team’s 150 genetic variants is present among Icelandic centenarians, he said.
This is more interesting stuff:
Are men necessary?
It seems to me that they exclude the main living arrangement in the west, that of a family consisting of a mother and a father and children.
I guess that was the only way they could conclude that Lesbian families were the best ... only compare them with single mothers, single fathers and lesbian mothers.
Why do I want to live that long? Life isn't very pleasant at more than half that age. I don't see it getting more pleasant at 80 or 90.
You remember homework assignments from almost 30 years ago?
From a biological perspective, longevity is not what counts but reproductive success (roughly, having as many viable offspring as possible).
Building an organism to last as long as possible must be balanced with that organism succeeding in the competition for mates and matings (i.e. sexual selection). Especially in males, where (in most species) the competition is greater.
If you are not worried about reproductive success, it has been known for many centuries how men can increase their average life expectancy - i.e. become a eunuch before puberty - e.g.
http://en.wikipedia.org/wiki/Narses
i worked with a girl from Sardinia (longest lifespans in the world)
It must be the sardines?
My Dad died at 53 and my mother at 91 hmmmm
This study was done with mostly Ashkenazi Jews in New York City, which may be related to the differences with the Icelanders (although a weak study design seems like a very plausible explanation).
"Very interesting beginning with an analysis of the service records of WW I veterans found that men who had at least four children before they were 30 were far more likely to live to 100."
I would be willing to bet a few nickels that that pattern reversed itself over the 60 years following World War I.
Seems to me that at the turn of the last century, having 4 kids before age 30 shows that you have your (*^ together. i.e. you can attract a mate; support a family; and you are healthy enough to repeatedly knock somebody up.
The cited 77 percent accuracy for predicting whether someone lives over 100, even assuming it wasn't biased by the data dredging you mentioned, doesn't tell you much about prediction in a case-control study.
If I found 10,000 on the street, and predicted that each of them would not live over 100, my accuracy would be well over 77 percent. Making predictions has to take into account the prior information that most people don't live that long. In a case-control study, classifiers are tricked into thinking this prior probability is 1/2.
I heard about this on the radio this morning and had more or less the exact same thought. Does your model predict 77% within the group you used to make the model? That would not be so impressive.
>The key to longevity is choosing one's parents well.< It helps if they ate your grandparents.
http://tinyurl.com/ycq4gd4
We have good evidence that people don't want to live that long. In fact normal sensible people don't care about longevity very much at all.
Consider ancient Rome. The best estimates are that the life expectancy was about 22. Those who lived in rural areas seem to have lived much longer. Yet this pattern persisted everywhere. It was still true in seventeenth century London - life expectancy around 25.
Cities have always been attractive to people but it costs them their lives.
It is ridiculously easy to get young men to go into battle or join a street gang or partake in dangerous activities. For example when I was twenty I sailed a boat through a hurricane, at night, while drunk. I had never sailed anything before. This was not the action of someone who was very interested in longevity.
Smoking costs you about ten years of life. Barrack Obama doesn't seem to care. The male homosexual lifestyle costs about twenty years yet young gay men still come to San Francisco. From an outside perspective they seem to be going there to die. From their perspective they are going there to live.
I have written down my life goals at several points throughout my life. I can't remember now every goal I set in my twenties but it certainly didn't include - live to be a hundred. The only people who seem to care about getting to be 100 are those who are 99.
Albertosaurus
There's no reason to use the one half division of in-sample out-sample. In fact usually out-sample can be smaller because you're only testing a single variable (the prediction from your model) instead of fitting over many degrees of freedom in the in-sample.
-Wall Street Quant
"Raising kids is stressful."
Depends on the kids.
My kids are a joy.
"disability-free at the average age of 93"
Few people would describe a 90 year old as disability free. In the future gene therapy may relieve many of the debilitating effects of aging (muscle loss, for example) but unfortunately most of the people reading this probably won't live to see it.
Whites live a long time but produce few kids. Blacks live shorter but leave behind many kids. I think the latter will win.
Seems to me that at the turn of the last century, having 4 kids before age 30 shows that you have your (*^ together. i.e. you can attract a mate; support a family; and you are healthy enough to repeatedly knock somebody up.
In some ways people were less fussy back then about who they married even though they should have been more fussy - rare divorce, no alimoney or child support, no government social services.
Personality mattered more back then. It was everything. Married couples were "role mates" and complemented each other. Geeks worked as office clerks and married empathetic nurses.
Now, it's Barbie meets Ken and finds each other to be "soul mates" and live happily ever after. And if you're not Barbie or Ken, tough luck. Geeks go into IT professions and meet other geeks while working their asses off. They marry at 30, have children at 35, and they turn out autistic.
There was a time when women needed to be gold-diggers to survive. Now liberated women with unrestricted careers and government support turn out to be even worse gold diggers.
Another interesting fact to emerge from centenarian studies is that Blacks are more likely than Whites to live to be 100. This is rather surprising, since Blacks overall have shorter life expectancies that Whites.
Blacks seem to have shorter life expectancies than Whites until about 80 years of age, after which the pattern is reversed, with Blacks having greater life expectancies than Whites the same age.
This stuff depresses me because 3 of my 4 grandparents died before I was born, ans the 4th died of
Alzheimer's in her sixties. My parents died in their mid-seventies. I'm probably outta here pretty soon!
I wonder if the NYSE volume phenomenon still holds, now that all World Series games are at night.
In the twenties, the Yankess didn't start their weekday games till late afternoon, because that's they way Wall Streeters wanted it.
Genes may have something to do with the aging but culture and diet play a much more crucial role. The Okinawans have one of the highest centenarian populations but when they leave the island and adopt a western diet or the mainland Japanese lifestyle of high fat high protein and high stress, their life expectancy is cut and turns out to be the same as everyone elses.
You should have also split the sample when examining over 300 correlations between state demographics and GOP vote share. With that many variables, there was bound to be junk findings.
Post a Comment