Doubly Robust Proximal Causal Learning for Continuous Treatments