If the AFM calculations are "predicated on a failure 1 second before V1", then surely that equates to roughly 5-10kts.
I know I wouldn't be able to make the decision AND first action within 10 kts of a rapidly approaching V1.
So the safest option has to be to continue if you have a failure just below V1.