Metodología y representación efectiva del conocimiento de guías clínicas para la ayuda a la decisión y protocolos clínicos

Abstract

1 Goals In our thesis we work with the hypothesis that the methodological assessment/selection of computable knowledge formalisms for Antibiotic Clinical Guidelines is essential for their effective use in Clinical Decision Support Systems (CDSSs) and Clinical Pathways (CPs), and therefore, for the standardization of clinical practise. For that, we pursue the following goals: • Goal 1: Define a framework of criteria to guide the selection of rule-based formalisms for the computable representation of Infection Guidelines knowledge, for its effective use in Clinical Decision Support Systems. • Goal 2: Prove that rule-based formalisms can be an effective way of representing infection knowledge of Clinical Guidelines, as a fundamental part of Clinical Decision Support Systems. • Goal 3: Define a set of key criteria to represent Clinical Pathways, analyzing scientific literature, BPM successive extensions to adapt to clinical settings and CG knowledge. • Goal 4: Prove that Infection CPs can be better represented using clinical-oriented formalisms, as a way to standardize clinical practise in infection treatment. • Goal 5: Propose extensions of the clinical-oriented formalism, in case gaps exist between the capabilities required for the representation of Infection CPs and the formalism’s native capabilities. 2 Methodology The framework of criteria of Goal 1 is based on the analysis of scientific literature and the knowledge expressed in the Antibiotic Guidelines of the John Hopkins Hospital. The validity of the proposed framework is assessed by means of a case study, comparing the knowledge representation process of the Urinary Tract Infection (UTI) using different rule-based technologies, thereby implicitly proving Goal 2. To reach Goals 3 and 4, we show how Clinical Pathways, often represented using general-purpose BPM notations, have concrete requirements for clinical settings, and more specifically, for infection treatment. To this end, we review scientific literature and analyze the peculiarities of Infection Pathways, representing the Catheter-Related Blood Stream Infection (CR-BSI) of the Antibiotic Guidelines of the John Hopkins Hospital, using the clinical-oriented formalism openEHR Task Planning. We also analyze the convenience of extending the used formalism to better represent Infection pathways, making an extension proposal, to cover Goal 5. 3 Results This thesis contributes to the definition of computational mechanisms for an improved computable representation of Antibiotic Guidelines in Clinical Decision Support Systems and Clinical Pathways, in two ways: • First, by proposing the use of rule-based formalisms as an effective way of representing expert knowledge in computable format within Clinical Decision Support Systems, and by defining a framework of criteria to guide the selection of a knowledge formalism, validated by means of a case study of the Urinary Tract Infection. • Second, by proposing a set of key criteria for the representation of Clinical Pathways, analyzing the native adequacy of the clinical-oriented openEHR Task Planning for the representation of infection CPs, using as case study the Catheter-Related Blood Stream Infection, and proposing extensions of openEHR Task Planning, for further improvements of the standard for representation of infection CPs. The results prove our hypothesis that the methodological assessment/selection of computable knowledge formalisms for the representation of Antibiotic Clinical Guidelines is essential for their effective use in Clinical Decision Support Systems and Clinical Pathways, thereby contributing to the standardization of clinical practice in infection treatment.