I am considering using causal random forests to detect/quantify potential causal effects of a variable of interest (e.g. radiation dose) on an outcome variable (e.g. some type of radiation-induced damage metric) in data sets where many other potential predictor variables are also present, and some of them can be strongly correlated with each other. Are causal random forests appropriate for analyzing such data? Can they apply only to controlled experiments (e.g. where the radiation dose was randomly assigned to subjects), or also to observational studies (e.g. where the dose value depended on some properties of the subject and/or other factors)? Thank you in advance for your interest and feedback.