h a l f b a k e r y
Good ideas at the time.
idea:
add, search, annotate, link, view, overview, recent, by name, best, random
meta:
news, help, about, links, report a problem
account:
Browse anonymously,
or get an account
and write.
Login
Create account.
|
|
| Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
| |
Can you cite the journal article
regarding
the predictability of sudden death by
MI? |
|
| |
Also, be aware that a test for any
disease is counterproductive unless it
has a low false-positive rate. Imagine
that your device predicts infarction with
a fale-negative rate of only 1% (which
would be astounding), but also has a
false-positive rate of only 1% (equally
astounding). |
|
| |
Now let's assume that there are 150
million Americans who get into their
cars twice a day (ie, 300 million "tests"
per day). Of these people, about 1000
will have an ischaemic attack at some
point in the day. On this basis: |
|
| |
a) Number of correct predictions =
990
b) Number of false negatives (failed to
predict MI) = 10
c) Number of false positives (wrongly
predicted an MI) = 3,000,000 |
|
| |
In other words, even with a sensor that
is 99% accurate both ways, you will
generate three MILLION false alarms per
day. |
|
| |
To put it another way, only one in 3000
warnings will be 'valid'. And this is on
the assumption of near-perfect sensors. |
|
| |
You really didn't think this one though,
did you, Treoand? |
|
| |
Umm, Max, if the false positive, and false negative for this test are both 1%, then I'm pretty sure that only 2970 in 3000 warnings would be "valid." |
|
| |
While this system is still going to produce a lot of false positives, I think Treon isn't the only one that needs to think this through again. |
|
| |
[ye_river] - you said 2970 out of 3000, [Max] said 990 out of 1000 - it's the same proportion of correct diagnoses. The important bit is that if the alarm goes off, (red flashing lights, a loud "You Are Having A Heart Attack!" voice, etc.) you can rest assured that there's a probability of only 990/3000000 that you really are having a heart attack (or to put it another way, 0.03% of the time the alarm goes off, it'll be correct). |
|
| |
You could reduce the incidence of false positives by making the alarm loud and sudden enough. |
|
| |
Ian, get thee to a BBC comedy scriptwriting
contract... |
|
| |
nfarct this, nfarct that. No nfarction idea here. |
|
| |
Those most at risk - the overweight, stressed, elderly, addicted and afflicted - would, I suggest, tend to register bloodflow irregularities from the simple act of levering themselves into a car. If they won't listen to their doctors, their families, their teachers, or Oprah, why will they pay any attention to a steering wheel cover? |
|
| |
Coming soon, the defibrillating airbag. |
|
| |
// If they won't listen to their doctors, their families, their teachers, or Oprah, why will they pay any attention to a steering wheel cover? // |
|
| |
Because for many people, their cars are very important. I think there's a good chance they'd listen to their car when nothing else could reach them, actually. |
|
| |