A thorough answer to this question will go beyond the scope of this forum. I dedicate three entire chapters in my book to the preparation, procedure, and implementation of all different types of FMEAs, including System FMEA. Having said that let me provide a few tips to get you started.
Let’s assume you know how to prepare and conduct an FMEA and want to know what is different about a System FMEA (including the interactions with users and patients). One of the first considerations is whether to include the interfaces with humans in the system hierarchy (as you have done), or to include the human interfaces as interface functions in the list of primary functions at the system level. I typically use the latter, but it can be done either way.
The most important thing is to include system interactions with humans somewhere, so they do not get missed in your FMEAs. If you include the interface between users/patients and the medical device in the system hierarchy, then it should be so described in the system hierarchy as “user-to-system interface” and “patient-to-system interface”, rather than merely “user(s)” and “patient(s).” If you include the user/patient interface to system as interface functions (in the FMEA function column, make sure they are properly described, including the standard of performance for each function. When including interfaces as functions, I like to state “Interface A – B” followed by the verbiage for the function.
It is also important to do an FMEA Block Diagram that clearly shows the scope of your FMEA, and shows the 4 types of interfaces (physical connection, material exchange, energy transfer and/or data exchange). One other point is to include in your System FMEA the interfaces between the various subsystems, as these can generate a significant percentage of the failure modes.
You will have to review your risk ranking scales to be sure they apply adequately to System FMEAs, Application FMEAs (if you do them separately), and Design FMEAs (typically done at the subsystem or component level).