3 ms·
I'm still curious what would happen if we: a) have a pill for every disease b) give placebos if we don't have anything better c) A/B-test viable new alternat
by mtrimpe 14y ago
I'm still curious what would happen if we:
a) have a pill for every disease
b) give placebos if we don't have anything better
c) A/B-test viable new alternative drugs (or maybe multi armed bandit's minimum regret is appropriate here ;)
That way patients might get the benefits of placebo without resorting to alternative medicine and you would be able to test effectiveness of drugs much more accurately.
You could even offer people different tracks, such as 'conservative', 'regular', 'progressive' and 'experimental' where different (new) drugs with different risk profiles can be issued.
If you'd make regularly contributing data (completing questionnaires) on your progress / health status a requirement for participation we could even start doing big-data analysis on this (massively valuable) data set quite soon.
- hamax 14y agoThe biggest problem A/B testing is ethical. Every patient deserves the best medicine we can provide, so A/B testing, even though it would work in the long run, is unethical.
- DanBC 14y agoWhen ethics comittees kill (http://www.badscience.net/2011/03/when-ethics-committees-kill/ http://www.badscience.net/2011/03/when-ethics-committees-kil...) > At present there is a bizarre paradox in medicine. When there is no evidence on which treatment is best, out of two available options, then you can choose one randomly at will, on a whim, in clinic, and be subject to no special safeguards. If, however, you decide to formally randomise in the same situation, and so generate new knowledge to improve treatments now and in the future, then suddenly a world of administrative obstruction opens up before you.