6 ms·

> For most tests, the risks associated with a false positive outweigh the benefits of marginally earlier detection. If these tests aren't run, then I am not su

> For most tests, the risks associated with a false positive outweigh the benefits of marginally earlier detection. If these tests aren't run, then I am not sure if it would be "marginally earlier", it might be much later or even never that the problem is discovered.

plonk 5y ago
If they are run, the results could be uninformative (https://en.wikipedia.org/wiki/Base_rate_fallacy https://en.wikipedia.org/wiki/Base_rate_fallacy). Too many false positives could lead to risky treatment and be dangerous. Testing each positive again would still create false positives. I don't know the numbers in this case, but I wouldn't be surprised if accurate mass testing were impractically expensive.
- closedloop129 5y ago
  This should only be a problem if you don't adjust the models. If you run a huge number of tests, you can create far more detailed models and thus reduce the false positive rate. Ideally, all chemical processes in the body could be known and we could have an almost "mechanical" model of the body. How else but with more testing can be reach that knowledge?
  - plonk 5y ago
    > This should only be a problem if you don't adjust the models. If you run a huge number of tests, you can create far more detailed models and thus reduce the false positive rate. Can you? The problem is the test's sensitivity and specificity, there's not much statistical work to do after the test on a single case. It's not an advanced model deciding whether you're sick or not. > Ideally, all chemical processes in the body could be known and we could have an almost "mechanical" model of the body. How else but with more testing can be reach that knowledge? Yeah, ideally we would. Not a doctor, but as far as I can tell, we're a thousand years away from that.
wirelesspotat 5y ago
The video in this comment provides some good examples for why the risks associated with a false positive can be pretty significant: https://news.ycombinator.com/item?id=31126368 https://news.ycombinator.com/item?id=31126368

plonk 5y ago

If they are run, the results could be uninformative (https://en.wikipedia.org/wiki/Base_rate_fallacy https://en.wikipedia.org/wiki/Base_rate_fallacy). Too many false positives could lead to risky treatment and be dangerous. Testing each positive again would still create false positives. I don't know the numbers in this case, but I wouldn't be surprised if accurate mass testing were impractically expensive.

closedloop129 5y ago
This should only be a problem if you don't adjust the models. If you run a huge number of tests, you can create far more detailed models and thus reduce the false positive rate. Ideally, all chemical processes in the body could be known and we could have an almost "mechanical" model of the body. How else but with more testing can be reach that knowledge?
- plonk 5y ago
  > This should only be a problem if you don't adjust the models. If you run a huge number of tests, you can create far more detailed models and thus reduce the false positive rate. Can you? The problem is the test's sensitivity and specificity, there's not much statistical work to do after the test on a single case. It's not an advanced model deciding whether you're sick or not. > Ideally, all chemical processes in the body could be known and we could have an almost "mechanical" model of the body. How else but with more testing can be reach that knowledge? Yeah, ideally we would. Not a doctor, but as far as I can tell, we're a thousand years away from that.

wirelesspotat 5y ago

The video in this comment provides some good examples for why the risks associated with a false positive can be pretty significant: https://news.ycombinator.com/item?id=31126368 https://news.ycombinator.com/item?id=31126368