"didn't combine them (as they are calculated differently......I...

"didn't combine them (as they are calculated differently......I can't see how they can combine them????)"

They can be combined providing the following conditions are true:
1) The necessary data points and diagnostic test results were present in the original data to allow alignment with the FDA calculations.
2) The optimised algorithm was re-run across all cough recordings to produce new predictions.
3) The new data points, diagnoses and predictions were used uniformly across all participants in the study.

I have emailed Tony and asked whether these conditions are met. If so then @samcro can relax; the 81% (cough + symptoms) represents a uniform application across all cases and there is no "91% + 59%" issue.

I'll let you know what the answer is when it comes. The timing might depend on where in the world he is at present.