I agree with what
@chris1price says, except for the part about product being worst case. For design validation, the units should be "representative product" (to use the term from 13485). Design verification should have already demonstrated that the design inputs are satisfied by the design outputs, including when those design outputs are at the limits of their acceptable ranges. The purpose of design validation is to show that the design meets user needs. For that, I would nominal units (i.e., any units that pass their final manufacturing inspections).
Now,
@caloykim2003, you mention you work for a software medical device company. I am going to assume that means your product is software-as-a-medical device. If you have not yet done so, you should check out the
IMDRF guidance documents related to SaMD. In particular, you should review
Software as a Medical Device (SaMD): Clinical Evaluation.
This guidance provides excellent information on the different facets of SaMD
clinical evaluation, of which
design validation is only a portion.
At the risk of oversimplifying things, as part of that
clinical evaluation, you will need to demonstrate that:
- There is scientific validity between the outputs of your SaMD and the healthcare situation or condition it is meant to address,
- Your SaMD correctly and reliably processes input data and generates output data with the appropriate level of precision (accuracy, repeatability, and reproducibility), and
- The use of your SaMD results in meaningful, measurable, patient-relevant clinical outcome(s) related to the condition of interest.
#3 is a large part of your design validation effort. You won't get that by using internal resources to do the testing.
You should also review FDA's guidance document
Applying Human Factors and Usability Engineering to Medical Devices. While not unique to SaMD, there is good information in here about structuring your validation testing. The takeaway from this document, as it related to your question here, is that (1) not only do your validation testers need to be representative of the population of actual users, but (2) the environment in which you perform the testing
also needs to be representative of the actual use environment.
(Related to simulated use environment)
The conditions under which simulated-use testing is conducted should be sufficiently realistic so that the results of the testing are generalizable to actual use...
To the extent that environmental factors might affect users’ interactions with elements of the device user interface, they should be incorporated into the simulated use environment (e.g., dim lighting, multiple alarm conditions, distractions, and multi-tasking).
(Related to test participant selection)
The most important consideration for test participants in human factors validation testing is that they represent the population of intended users....
The number of test participants involved in human factors validation depends on the purpose of the test... in general, the minimum number of participants should be 15.
I know this is a lot, but hopefully it provides the clarity and direction you are seeking.
