So, according to this, the false positive rate for the Apple Watch in detecting atrial fibrillation is 0.04% (99.6% correct) This means that, on average, Apple Watches will be wrong more than 80% of the time Sound counterintuitive? This is the issue with population screeninghttps://twitter.com/statnews/status/1040273579391242242 …
-
-
Assuming that the false negative rate is 0% (this is pretty close, it was actually 2%), in a sample of 1000 people you'll get 10 with AF who will receive the correct diagnosis
Show this thread -
Now, the false negative rate is 0.04% So, in the remaining 990 people who are tested, 40 will receive positive results
Show this thread -
Keeping track, that means 4 out of every 5 people who are diagnosed using this app will be diagnosed incorrectly
Show this thread -
And remember, these people will then likely go on to receive unnecessary tests, interventions, and potentially medications Some of them will probably die
Show this thread -
This is why population screening is a fine art Even if your test is FANTASTIC, you have to be really damn careful Because otherwise you can KILL PEOPLE
Show this thread -
P.S. just in case this wasn't elucidated well above - 1% prevalence means that 10 in 1,000 people have the disease and AF=atrial fibrillation
Show this thread -
P.P.S - As has been pointed out, I got my simple maths incorrect here Serves me right for being overconfident! The true negative identification should be 4, thus the watch is only wrong ~1/3 of the time Still a fascinating phenomenon
Show this thread -
This false identification rate can be reduced if the proportion of the population at risk increases For example, if 10% of the population have AF, then the watch only gets it wrong a tiny percentage of the time
Show this thread -
This is why screening is usually only recommended in very high-risk populations
Show this thread
End of conversation
New conversation -
-
-
Whats population prevalence of AF among iWatch users though... I wonder if Apple has basic demographics
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.