Generally:
It is always a good idea to have a process compliant with 14971. The difficult truth is that a third-party will almost always find something wrong with your particular implementation, but if you have been
trying to implement a 14971-compliant process you will ultimately be better off should it come time to 'remediate' (depending on your attitude and the attitude of whoever establishes your remediation plan). You need a 14971-compliant process to motivate 62304.
62304 is ultimately a 'best practices' process standard that establishes the minimum necessary development deliverables, based on a risk categorization. Many professional software developers who take pride in their work will do the full suite of activities
anyway, no matter the risk categorization.
The usability standards
minimally ought to be considered for the validation of the SaMD; i.e. is the SaMD meeting the needs of the users. Frankly: when it comes to SaMD there is always going to be a learning curve and Luddites will be pre-disposed to not like all sorts of things about any SaMD. Focusing on colors, fonts, displayed text (and languages), UI layouts would probably be enough to satisfy most third parties... the things you focus on should be motivated by the initial risk assessment.
My own feeling is that
usability of SaMD is going to take a backseat to
security of SaMD for the foreseeable future. The usability standards and guidances are still important and serve an important role in the design of medical devices, of course. My thinking is motivated by:
- Software design space (including User Interfaces) is simply too "wide open" for most companies to mandate/precisely specify UI
- The hand-held computing revolution has (a) exposed many more people to different UI and (b) established "islands of stability" for acceptable UI (that is, there are fewer Luddites and a wider universe of recognizable user interfaces)
- Security issues around integrity and availability will almost always trump particular UI design choices (made primarily around the concept of 'essential performance') in a risk-based analysis.
