Shap Values for uplift

[MarioDamhur]

Is it possible to identify the features that most impact the "sensitivity" of treatment with SHAP values?
If I use, for example, the ClassTransformation model for uplift modeling, when I try to use the shap.Explainer(best_model.estimator), it requests the base estimator objetct and not the ClassTransformation object. So I think the SHAP values give the impact on the probability of the label and not the uplift prediction for the model. Am I right?