Given high-dimensional data describing differences in characteristics between individuals, what state-of-the-art ML technique is best suited for estimating differences in causal effects?