3 ms·

96% accuracy may be a problem... Depends how many false positives and false negatives, but if there's 4% of false negatives and there's, say, 1% of the tested

96% accuracy may be a problem... Depends how many false positives and false negatives, but if there's 4% of false negatives and there's, say, 1% of the tested population that has the disease, then even a positive test may mean you have only 20% chances of being actually positive. (someone double check my math, but the general idea stands, I think)

corin_ 6y ago
The actual data is more complicated than "96% accurate", but I'm not the best person to deep dive on the numbers and give a better analysis. However from reading the abstract of the actual study, it seems the TLDR is "more accurate than current ways to test, and significantly cheaper too" - which would be a great step regardless of if it's not quite as perfect as a vague "high % accurate!" makes it sound. https://jamanetwork.com/journals/jama/fullarticle/2768841 https://jamanetwork.com/journals/jama/fullarticle/2768841