A Marriage Of Machine Learning Methods And Causal Inference Parameters To Derive Counterfactual Dose-response Curves For Air Pollution And Respiratory Outcomes