I think there's merit in the idea -- not tracking what a code reviewer reads but more of tracking how many times each line of code has been _included_ in a review, by being modified or maybe within so many lines of a change (like how diffs show X lines of surrounding context). The idea would be that code changes _near_ a buggy line would be more likely to draw attention to that bug and perhaps lines with less attention would be more likely to facilitate hidden bugs.
I suspect it would work out the other way round - code in a frequently modified section would be more likely to hold bugs because the requirements/understanding of that code is changing more frequently.
