4 ms·
Not even close. Suppose you're testing for something that only occurs in 1% of the population or less. What if the Theranos test just never returns a positive?
by landryraccoon 11y ago
Not even close. Suppose you're testing for something that only occurs in 1% of the population or less. What if the Theranos test just never returns a positive? Then that test is "99% accurate" compared to the "normal" test.
Of course this is just a hypothetical, in real life you have to do a carefully controlled study - which is exactly what the FDA requires.
- w1ntermute 11y ago> Suppose you're testing for something that only occurs in 1% of the population or less. What if the Theranos test just never returns a positive? Then that test is "99% accurate" compared to the "normal" test. Understanding statistics is very important here: http://www.kevinmd.com/blog/2015/11/the-controversy-over-theranos-misses-the-larger-point.html http://www.kevinmd.com/blog/2015/11/the-controversy-over-the...
- SilasX 11y agoPerhaps if it's a binary test, but blood tests typically return a cardinal/continuous value; it's enough to verify that e.g. Theranos returns something close to 44 when the reference test returns 44. No one's disputing that bad statistics can invalidate some comparisons.