You're right - short of diving in deep on the backend, there doesn't seem to be a good way to do that.
If you've been monitoring these devices for awhile, however, you do have some baseline measurements for what 'normal' is. Perhaps you could simply modify the specific hardware sensor's thresholds to trigger on 'normal avg + x (where x is approximately 10% of the norm reading)' for warning and 'normal avg + y (where y is approx. 15% of said reading)' for critical.
If you've already investigated this and found it insufficient, then forgive the duplicate info.
If you DO find out how to alert on 'avg + 10%' on these sensors natively, please share post-success - I'm sure others would like to have a ready way to do this.
I believe this is the functionality you are looking for, or at least the closest existing functionality to what you are requesting: https://support.solarwinds.com/Success_Center/Network_Performance_Monitor_(NPM)/NPM_Administrator_Guide/120_Thresholds/040_Baselines_and_baseline_calculations