MR-MoE uses machine learning to choose the best performing model out of multiple pleiotropy-robust estimation methods.
Based on meta-learning, in which simulations are used to train a machine which selects the method most likely to return a reliable MR estimate.