Predictions and Perfect Science
If a person about to act was aware of the prediction, I believe the prediction would still be accurate if the person did not act on a “learned helplessness” (for lack of knowing the proper term) of sorts. Meaning, it could be accurate if the person aware did not purposely change their predicted act to disprove the prediction.
The fact that a prediction can be known in advanced may not always disprove that it can be accurately predicted, so long as the person about to act is unaware of the prediction. If they are aware, then it is more likely that the accuracy will be discredited since the person can intentionally act in favor of the prediction.
These predictions can be inaccurate either way, I think, if only one prediction is made when there are a number for a specific action. I find it inaccurate to say a person will cover their mouth when they sneeze, yet there are other predictions such as the person not covering their mouth or they might prevent the sneeze entirely when they feel it coming on (press the nostril you feel is more bothered before sneezing, works most times!).
There are only two ways I see a single predicted action being accurate. One is if a person and their actions are well known so it may be predicted that this person will reach in their pocket/purse/etc. and pull out a tissue to cover their mouth when sneezing. At this point, detailed predictions could also be made like the person will try to ease the sneeze, fail, and quickly cover their mouth in their elbow to sneeze. The second way is if there was only one possible outcome for an action. But I cannot think of anything with only one possible outcome at the moment to give a better example.