Usefulness of triglyceride-glucose index and homeostatic model assessment for predicting coronary microvascular dysfunction.