How framing worklist component selection as a Learning-to-Rank problem reduced algorithmic iterations by 35.8%.