Re:What Killed My Sister?
by pdpi on Tuesday April 29, 2014 @07:17AM
(#7666193)
Depending on what they're measuring, 98.3% can be pretty bad. Given that there's about a 1% portion of the population with schizophrenia, if the test is right 98% of the time (irrespectively of everything else), then it'll misdiagnose two healthy people as having schizophrenia for every correct diagnosis.
[ Parent
| on HN ]
Yes, but wouldn't it be obvious (or obvious eventually) that these healthy people clearly don't exhibit other symptoms of schizophrenia? Or am I forgetting something?
What about terrifying people with the thought they have schizophrenia but just haven't demonstrated visible symptoms yet? That concern comes up a lot with genetic testing because many things aren't certain but people panic and make significant life-altering decisions after a positive test result.
EDIT: … I should also have noted that humans are notoriously suggestible – with a mental illness it's really easy to imagine people both assuming every quirk is proof of a terrible disease or even starting to demonstrate symptoms psychosomatically after being told that this will happen eventually.
Good point. I suppose it should be limited to people who are already showing other symptoms of schizophrenia...
… or with family histories, etc. The real challenge, though, is probably just for doctors to be very open about false positive and negative rates and explain the issue directly. I think a lot of these issues would be easier if someone was frankly told “3 times more people receive positives on this test than actually have the disease“ instead of “it came back positive for [scary]”.
Nevermind the classic problem that was addressed in One Flew Over the Cuckoo's Nest and others: how do you prove you're sane? If the machine says you're insane, what way is there to prove otherwise?