Medical devices are far more loosely regulated than pharma, and IMO a lot more innovative (maybe as a result of shorter development/trials cycle, easier testing, and thus lower capital requirements). You literally have hobbyists, garage entrepreneurs, etc. still, which is amazing (and a huge Israeli contribution).
Medical device development is so interdisciplinary that I have zero confidence in the government being able to staff/grant/etc. its development. The only people in government likely to create medical devices would be end users who happen to be hackers, or government employees (at places like NIST) who are technical, need the product personally, and end up developing it on their own as a side project.
Government-funded R&D might work for NIH style long term research which is very structured and well understood, but I think it would be horrible for "hacking", which is what medical device creation largely is. What is really needed for devices is a bunch of bootcamp type programs for non-biomedical engineers to learn about the process, and clinicians posting to the Internet with their problems with current systems. I think today one of the main ways developers find out is when they or a family member/friend goes in for a procedure and they see a need.