Confidence interval for Diagnostic Accuracy (sensitivity and specificity)

Emmanuel_Romeu

Hello,

Medical devices company oftenly use Diagnostic Accuracy (sensitivity and specificity) to evaluate their devices.

During the R&D process and later at V&V (verification and validation) they have to provide the confidence interval using the Wislon score method 

https://www.itl.nist.gov/div898/handbook/prc/section2/prc241.htm

as recomended by FDA.

An add-in Performance Summary Add-in currently exist that provide such CI however a hard copy would be beneficial as regulatory bodies will not accept a result from an add-in without extensive validation following 21 CFR part 11.

 

Emmanuel (former  Beckman-Coulter employee)

 

karencopeland

I wrote this add-in a number of years ago now as a solution to a specific set of measure that I needed frequently and could only get out of the distribution platform (utilizing a table of counts with frequencies). Adding the ability to get performance CI metrics from the new decision thresholds would be one place to add in CI.  A more general location would be in contingency tables. It would take some thought as to how to implement...row CI, Column CI, overall CI?? I would want the score CI as in distribution.