When is a medical “breakthrough” never a real breakthrough?
Nearly always.
Simple 2 reasons: first, these days medical researchers hype their
data beyond all reason, and second, very, very, very few medical reporters have
absolutely any idea at all about how to interpret what they are fed by PR
departments, on top of which hardly any of them – us, really – know the first
thing about statistics.
Great example: a study out this week heavily hyped by British
reporters and headlines (researchers were British) about a “breakthrough” blood
test for Alzheimer’s in which the researchers claim to have come up with a test
that can determine with “up to 87 %” accuracy which of the many people who have
mild cognitive impairment (MCI) will actually go on to get AD over the next 10
years.
Only trouble is, when people better at this than me parsed the data, they
concluded that the real predictability of this test as to whether any
particular individual may develop AD is roughly 50 % (in fact, it can be as low
as 47 %).
Hey, I don’t have to know anything at all about you and I can predict
with 50 % accuracy whether you – or I – will develop AD simply by guessing
“yes” or “no”.
Doubt that that would qualify as a breakthrough, though.
