PDA

View Full Version : Medical science -- genetics question



douglass
06-06-2012, 07:50 PM
Hi,

I asked this question on a science forum and so far have had one response and thought I'd ask here too.

My novel is set in the future and wanted to find an opinion on the future possibilty of determining a person's life expectancy down to a fairly specific prediction. For example, If you had a very old man and access to all of his medical information (dna, etc.), could you predict he had 24 months to live, plus or minus one month? (assume death by natural causes -- no cancer, accidents, etc.)

Would that ever be possible?

If so:
What sort of medical info from the person would be needed?
What science breakthroughs would be needed?

Thanks in advance for any opinions.

Piddlepup
06-06-2012, 08:30 PM
I do databases for a living as well as studied genetics for awhile so i guess I can take a wack at this question.

Would it be possible? I believe so.

You would need a complete DNA map from the person and as much additional data as you can get.
For example, their parents DNA, their grandparents DNA, DNA samples from where they live (country, state, county, state). The more DNA samples the better.

What science breakthrough would be needed?
You would need as much data as can humanly be generated.

The idea you should be going for is that the more data we can find on a particular marker in their DNA the more likely you can predict if that marker is better or worse for someone. And the more data on how long people with or without that marker live will give you. This is basically creating a statistical database and generating odds.

We do this all the time in the business world and I'm sure science is doing it now.

Torgo
06-06-2012, 08:38 PM
I seem to remember people were concerned enough about this possibility that they have already passed laws in many countries to prevent people from being discriminated against by insurers using possible future tech. Here's a quick paper (http://www.guythomas.org.uk/pdf/GI.pdf) you might be interested in (ten seconds of Googling so sorry if it's not what you were after.)

veinglory
06-06-2012, 08:38 PM
It raised the question for me of why you could predict something so accurately and not prevent it.

Basically you used to just die of 'old age'. Increasingly we now know the proximate cause of death.

But if you know the time very closely, you must in order to predict that know the cause (e.g. aneurism, heart failure) and so have a chance to treat and so beat that date?

lorna_w
06-06-2012, 08:42 PM
I'd totally believe it in a story set in the very near future. You wouldn't even have to do a lot of explaining for me; I'd just accept and move along.

lbender
06-06-2012, 08:50 PM
I'd never believe it. There are too many environmental factors that have nothing to do with genetics. If you want to figure out approximate life spans, sure, but down to a month?

By the way, the term 'natural causes' is a catch-all phrase that doesn't mean much. It does eliminate getting hit by a car or a truck or a bullet, but that's about it.

Siri Kirpal
06-06-2012, 09:28 PM
Sat Nam! (Literally "Truth Name"--a Sikh greeting)

I don't think you could get that accurate with just DNA. You'd also need to know the environment and lifestyle habits. A smoking meat-eating guy and his vegetarian non-smoking identical twin won't have the same natural life expectancies. And neither will identical twins where one lives near a chemical dump and the other lives in a more healthy environment.

But factor in those things, and yeah, I think it could be done.

Blessings,

Siri Kirpal

douglass
06-06-2012, 09:51 PM
I appreciate everyone's input.

I will add -- the environment is like a monastery.

Thanks again.

mirandashell
06-06-2012, 09:55 PM
I reckon you could get away with some handwavium in this. As a Sci-fi reader, I don't need to know every single detail of a technology. I just need it to read like it would work. Enough detail to make it plausible.

Of course, where you draw the line is the tough question

jaksen
06-06-2012, 10:01 PM
If I wrote a story with this as a premise, or presumption, I'd exploit it. I'd have people learning this information and using it against others, or if a person learns it's nearly inevitable he'll die before, say, fifty, it's a major part of the story. He either struggles to beat the odds, or becomes an enlightened soul doing great deeds or if he's of a different bent, he lives a risky, wild and carefree life til the end.

But then, what if at fifty - after he's used up all his money, friends and family - the doctors say ahhh, we made a mistake. You have at least fifty years to go?

The idea itself gives way to endless story ideas, imo.

onesecondglance
06-06-2012, 11:09 PM
I used to work in annuities, which are a UK-specific whole-of-life financial product. How good your company is at making money out of annuities depends largely on how good you are at estimating mortality rates - working out when you're likely to die.

With that background knowledge I wouldn't accept this in anything but a soft SF or far-future setting. There are just too many environmental factors that aren't genetic and the risk of a lethal infection just popping up - especially in a closed monastery environment - make any guess at age from cells implausible to me. And even then sometimes people just die when they seem perfectly healthy. Or live when they seem really close to death.

It's a nice idea though :)

lac582
06-07-2012, 04:48 AM
Natural causes is kind of a funny term these days, isn't it? If we mean organs shutting down with no specific cause, then I think it might be far-fetched to predict that. I could buy predicting when the kidneys should shut down if someone had an underlying chronic condition and wasn't treating it with dialysis, though.

You could look into Kleiber's Law and the metabolic rate of animals. If in your future medicine allows really individualized assessments of metabolism, then perhaps they could make predictions based on that.

See this exchange (I make no claims on its accuracy): http://skeptics.stackexchange.com/questions/5701/does-every-species-get-around-a-billion-heartbeats-on-average

espresso5
06-07-2012, 08:49 PM
Actuarial scientists could do a pretty good job of predicting life expectancy based on hereditary and lifestyle information before the days of the human genome.
Now we have the human genome, which is a sequence map of all the genes in a human. One problem is that it is still expensive, about 5k per genome, but will soon be down to 1k. In the future you will probably be able to get a genome mapped for a hundred bucks in a week or less.
The bigger problem in using genetics to predict life expectancy is that there is still a considerable dearth of knowledge in how the genes interact. For instance, we know there are 20-25k genes in the genome, implying that we don't even know if 5k genes are even genes, much less what they do and how they interact with other genes.
Over the next few decades, using highthroughput technology, which allows scientists to look at the effects of multiple genes on cells and an organism at once, many of these relationships will be worked out.
In addition, we have to figure out how these genes not only interact with one another, but how they interact with the environment. Epigenetic changes seem to have a bigger effect than once thought.
Depending on how far in the future your story is set will determine how much of the genome and its interaction with the environment, and its overall effects on life expectancy will be known.
I would say in fifty years, though, using information from an individual's genome as well as their lifestyle choices (perhaps derived from some type of lifelong record from the future equivalent of facebook and twitter), someone should be able to predict death from natural causes with a degree as precise as you refered to (24months plus or minus a month).

douglass
06-07-2012, 10:13 PM
Thanks again, everyone!

Espresso -- thank you so much for the detailed response. That is a huge help to me. If you have a moment, could you give a brief explanation of "high throughput technology"? Is that having to do with very fast computer analysis? Again, thank you.

veinglory
06-07-2012, 10:17 PM
There is big difference between an actuarial table and predicting the life span of an individual--scientists up to the current day have very little chance of doing that accurately.

Izhitsa
06-07-2012, 10:42 PM
There is big difference between an actuarial table and predicting the life span of an individual--scientists up to the current day have very little chance of doing that accurately.

A deep actuarial study of an individual accounting for heredity and morbidity has better chances. It is known that Lyndon Johnson commissioned an actuarial study of himself, and the conclusion (he was likely to die during the second full term of presidency) was one of the factors which made him to forgo the '68 presidential bid. LBJ died a mere two days after the would-be second term ended.

veinglory
06-07-2012, 10:47 PM
Yes, and famous hits overshadow the vast majority of misses. Unless the person is in the throes of progressive disease, it's a crap shoot. LBJ had advanced heart disease, had experienced a near fatal heart attack in 55, and had not changed his lifestyle (high stress, poor diet, heavy smoker). Saying he would die within 5 years was not difficult. You would need to know any more than what it said on his medical record.

douglass
06-07-2012, 11:06 PM
I know a bit about actuarial statistics and that they offer a general clue.

Now, I'm interested in the future plausibility of using analysis on an individual genetic and cellular basis to make more accurate predictions.

Izhitsa
06-08-2012, 12:38 AM
Yes, and famous hits overshadow the vast majority of misses. Unless the person is in the throes of progressive disease, it's a crap shoot. LBJ had advanced heart disease, had experienced a near fatal heart attack in 55, and had not changed his lifestyle (high stress, poor diet, heavy smoker). Saying he would die within 5 years was not difficult. You would need to know any more than what it said on his medical record.
Yet, were some actuaries to make the same 5-year survival prognosis for Brezhnev after his '74 cerebrovascular accident, they would've been off by three years. The medical record is good, but adding family history might provide better results.

As for 'famous hits overshadow the vast majority of misses' - no doubt. Famous people get better actuaries, I suppose. :-)

espresso5
06-08-2012, 02:09 AM
Thanks again, everyone!

Espresso -- thank you so much for the detailed response. That is a huge help to me. If you have a moment, could you give a brief explanation of "high throughput technology"? Is that having to do with very fast computer analysis? Again, thank you.

Traditional geneticists and protein biochemists might spend a career on a given gene, the protein for which it codes, and the effects a mutation in that gene might have on an organism. Many of the molecular geneticists trained in the eighties and early nineties still use this approach. Just for the record I'm not implying there's anything wrong with this, and I appreciate that this level of indepth study still has its uses.
Since the publication of the HGP ten years ago, newer technologies have developed that allow scientists to look at hundreds and even thousands of genes and proteins in a given system (cell, organelle, organism, etc.) at a given time. For instance, a cancerous cell might be over or under expressing different genes or sets of genes compared to a healthy cell.
By looking at many genes at once, we can hope to determine how these genes interact.
In the eighty and nineties, it was thought that if only we could sequence the human genome, we could revolutionize medicine and our understanding of how genes affect health. There were many examples of mutations in single genes having drastic effects on people, up to an including death. If we knew all the genes, think of what we could learn. Unfortunately, the single gene defects are largely the exception, and were discovered so readily because they were the low hanging fruit.
Most traits are polygenic and multifactorial, meaning there are many genes contributing to the trait and there is an environmental component. For instance, its thought that as many as five hundred genes contribute to intelligence, and of course factors such as education can determine how much mileage one gets from the raw intellect.
Genomics, proteomics, bioinformatics, epigenetics and the other new -ics should help answer many of the questions about how genes, proteins, and the environment work together to result in the final phenotype by analyzing vast amounts of information at a fraction of the cost and time that was required even a few years ago.
For instance, the first human genome took eleven + years to sequence and cost close to three billion dollars. By the early '00 the price was down to a hundred million dollars. By 2009 it was 100k and today it's down to less than 10k and a few weeks. The 1k genome is on the horizon, which will allow researchers to finally start getting lots of data on many individuals to compare the genomes and genes. In ten or twenty years, getting your genome sequenced will probably be as standard as is getting a CT scan today.

douglass
06-08-2012, 06:00 AM
Espresso,

Thanks so much for taking the time to write this detailed explanation!