RE: 99.99%9 Jun 2021 13:29
NDN71 your point about competitors is fair. I accept that
Beinthelead by significant I mean an accurate assessment. There is no way this test works 100/100 in the real world with all the potential variance in co dictionary, setup and users. Higher sample numbers give higher confidence that the results you are seeing on sensitivity and specificity are close to true performance. Lower boundary of the 95% confidence interval is a far more valuable number than the headline sens/spec.