Act as a Senior Data Scientist. You have been given a large observational dataset regarding patient outcomes across different hospitals. You need to determine the causal effect of a specific treatment protocol on patient recovery, despite the presence of confounding variables and selection bias (e.g., healthier patients being chosen for the treatment). Describe in detail which causal inference techniques you would apply (e.g., Propensity Score Matching, Instrumental Variables, Difference-in-Differences). Explain how you would validate the assumptions of your chosen model and quantify the uncertainty in your causal estimates.