摘要：Decision making trial and evaluation laboratorymethod and semantic structure analysis are the two mostpopular approaches to study structural analysis of criticalfactors or dimensions but do not have solid theoreticalbackgrounds to construct causal relationships among criticalfactors or dimensions. This study proposes a framework byusing normalized mutual information for both continuous anddiscrete random variables to set up the contextual relationshipsfrom the raw data. A Monte Carlo simulation based onnormalized mutual information is conducted to estimate thethreshold value of establishing the causal relationships. Twointernal survey data of safety attitudes questionnaire in terms ofthirty questions and six dimensions are used from viewpoints ofphysicians and nurses. Four cases are illustrated to show howthe proposed framework works in practice to identify criticalfactors or dimensions from the internal survey data. Thisproposed framework enables hospital management to initiateimprovements from causal factors or dimensions to effectivelyenhance patient safety from medical staffs’ viewpoints.
关键词：normalized mutual information; random;variable; Monte Carlo simulation; safety attitudes questionnaire;patient safety culture; causal relationship
