There are some things that feel like they should reduce the risk of RSI when developing a user interface, such as having more common menu options near the top of a menu to reduce the amount of mouse travel required, but are there specific guidelines (ideally backed with research) that can be used when designing the user interface of an application to reduce the risk of repetitive strain injuries in end users of software?
Look at your target users and see what the existing best practices are for ergonomic UIs. It is going to vary by industry. I'll give a specific example of a UI that we implemented to reduce RSI.
Our company introduced a new order management system about 18 months ago. Through our user research, we discovered that our power users often complained about the old UI, because it forced them to switch between keyboard and mouse frequently (click on a search field, type in a search term, click to search, type in quantity, click to accept, etc.). This work flow was fine for someone who only used the system occasionally and actually needed to search for the product they were ordering. But for our power users it was tedious and time-consuming.
One of our developers had experience with purchasing software in another industry and recommended that we create a number pad-only UI for power users. We defined these users as people who already knew our part numbers, regularly ordered long lists of parts, and used our system nearly every day. The power user UI allows users to enter as many parts as they want without moving their hand from their number pad. Type in part number, hit enter, type in quantity, hit enter, type in next part number, etc. In user testing, we discovered that our solution indeed worked for power users, and it's been adopted by most of our power user audience.
From the top of my mind, for traditional desktop interfaces, no research backing to list:
"the biggest button" a.k.a. desktop corners and edges (you can ram the mouse in at any speed
Sub menus open centered around the selected parent menu item, to reduce overall travel time
"magnetic" controls (like snapping edges when resizing or moving windows)
Making "everything" accessible by keyboard and by mouse, so users can switch
Macros / automation
I remember reading about some of them having research backing, but that memroy is really fading. The common themes are avoiding long travel distances and necessity of exact positioning, as well as giving the option of keyboard use.
Few of them carry over to touch interfaces, which I have too little experience with to speak up.