Eliciting patients' preferences for epilepsy diagnostics: A discrete choice experiment