Another way to do it with actual checkboxes/radio buttons:
In each row of your repeating table, have a radio button, RAD, and an unseen integer field, STATUSFIELD, and another integer field, WATCHER. Outside of your repeating table, an integer field CURRENTHIGH, set to 0.
When the user clicks a radio button, set current()/STATUSFIELD and CURRENTHIGH to CURRENTHIGH + 1. Make a rule for the field CURRENTHIGH, which will be "set off" when CURRENTHIGH changes. Make the rule set all occurrences of WATCHER to WATCHER + 1. This will change WATCHER. Now make a rule for the field WATCHER (this will happen now, because WATCHER has changed) to set current()/RAD to its "off" value if STATUSFIELD != CURRENTHIGH, meaning that it wasn't the last radio button clicked.
When you need to get the current value of RAD, calculate RAD[STATUSFIELD = CURRENTHIGH].
Does this makes sense?