It would appear that I could...
Click to view a video clip (1MB WMV)
There are two cables for the mouse as there were for the keyboard - one goes to the TI, the other goes to the power supply. The mouse is put into the lowest resolution mode (the cursor is moved 1 pixel per mm).
There's a demo for anyone with a TI-83 Plus and an adapter. It also contains a keyboard demo - and there's a video of that demo here (1MB WMV).
Fortunately, I didn't have to butcher that mouse (it's a rather good miniature USB/PS2 one), but it is connected to the TI with Blu-Tak to keep the wires on the contacts, and isn't very mobile as I only gave the power lead about 15cms clearance. I think it's high time I grabbed myself some proper connectors and a soldering iron from Maplin... [rolleyes]
As for the mouse routines - they need a lot of tidying up and making much more reliable, especially in the recovery when a mouse is disconnected then reconnected (my keyboard routines are 'hot pluggable', after all). Extending them to support the Microsoft Intellimouse system, giving the mouse an extra axis (scroll wheel) and two extra buttons would be rather cool as well.
Ultimately, these routines are rather useless until they become widely adopted. Ideally, they could/would be added to a shell, and form part of the standard routines. For example, the shell could provide a "get key" function, which would also transparently return keyboard keys as if they were keys on the keypad, or return four "left" keys if the mouse was moved 4 mm left -- that sort of thing.
Chances of this happening are, sadly, nil. [rolleyes]