Medical Device Design verification sample prototype

[Ronen E]

Thank you Ronen

Do you know how do we decide the number of times you can run 1 unit ?

It's an engineering decision IMO. I think that in most cases 2-3 times per unit would pass without question.

 

[Watchcat]

Unless it's a disposable device, you are generally allowed to run each unit multiple times to enlarge your sample size.

In that case, what is it you are sampling? It can't be units, because if you run the same unit run multiple times you have still sampled only one unit.

 

[Ronen E]

In that case, what is it you are sampling?

Uses.

 

[Watchcat]

So you are verifying that one unit will pass a test after multiple uses? Is this for a wear test?

Whatever unit(s) you use, I think that they will define acceptable tolerances, won't they? With one unit, then you have verified that units made to those exact specifications (no tolerances) will pass the test? If you use a few units then you have verified that units made within the range of tolerances represented by those few units will pass the test?

PS I'm not adding to this discussion, but pursuing Design Engineering for Dummies, so nobody needs to knock themselves out to enlighten me. I won't ever even do any of this stuff myself. Just have an inquiring mind that wants to know.

 

[Ronen E]

So you are verifying that one unit will pass a test after multiple uses? Is this for a wear test?

No. I wasn't answering from a reliability / life testing perspective.
I would never recommend testing just one unit (no matter how many uses) for verification, unless the whole planned production of that model was 5 units or less. Indicative testing of one unit can sometimes be useful in the development process, though - to get a feel whether one is in the right direction.

The general idea above was that when only a limited number of units is available for testing (say, 3-5), the available data set size can be increased through running (testing) each unit multiple times. Running the same unit multiple times doesn't address manufacturing-induced variability, but it does account for the "natural" sampling variability that will be present in real-life use anyway (i.e. if you run the same physical unit again and again under "the same" conditions and settings you will hardly ever get the same results repeatedly).