The most common mistake in the final week is spending most of it covering material you have not studied yet. The cost is double: the new content does not stick, and question types you had already mastered lose their edge because there was no time to keep them warm. A steadier approach is to change the week's goal from increasing coverage to reducing variance.
Sleep schedule comes before content. If exams start in the morning, shift your waking time and your peak-focus window gradually over the seven to ten days beforehand. Fixing your wake time is easier to sustain than forcing an earlier bedtime, and it moves your body clock faster. The extra hours gained by staying up are usually paid back with interest in lost concentration during the exam.
Practise under time pressure, and do it during the same part of the day as the actual exam. Treat each past paper as a sitting: stop the clock, no phone, no breaks beyond what the exam allows. After each paper the most useful output is not the score but the error classification — misread the question, did not understand the concept, or ran out of time. These three need completely different fixes; reviewing them together means reviewing nothing.
In the last two days, stop attempting new question types. Go through your error log, your own formula sheets, sentence patterns and the characters or spellings you consistently get wrong. Check the logistics once, properly: admission slip, exam venue and travel time, permitted items, identification. Mistakes here cost far more than skipping one topic, and they happen every year.
If one subject is clearly weak, this is not the week to reverse it. Secure the basic marks instead: drill the question types that come up most often and are easiest to score, until they are automatic. Reliable basic marks usually beat ten extra hours spent on hard questions you have never really learned.