The use of sliding time windows for the exploratory analysis of temporal effects of smoking histories on lung cancer risk