Proaktive Assistenz zur kontextabhängigen und zielorientierten Unterstützung bei der Indikationsstellung und Anwendung von Behandlungsmaßnahmen in der Intensivmedizin

During patient care physicians have to make important decisions, often based upon uncertain and incomplete data, about diagnostic and therapeutic actions. Various kinds of decision support exist in form of paper-based recommendations and guidelines, yet only a few proactive systems support the physician in selecting and executing these guidelines. This thesis contributes to the research on computerized clinical guidelines by assessing, combining and extending existing methods and technologies used for automating clinical guidelines:- A meta-model is developed to capture process knowledge which exists e.g. in form of guidelines. The model is based upon the main common elements of computerized guideline languages and is for the first time extended by a strategic level by means of specific goal modelling.- An execution engine is implemented which is based on flexible workflow techniques and connected by an HL7 interface to communicate with a patient data management system. User interaction is through the electronic patient record in the patient data management system only. Thus, the execution engine integrates into the working environment of a physician seamlessly.Thus, for the first time computerized guidelines are combined with explicit means of goal modelling and automation accompanied by an iterative modelling method. A tool chain for formalizing and automating clinical guidelines was implemented and evaluated with physicians of an intensive care unit in the context of a research project part-financed by the German Research Council (DFG). This paves the way towards the assessment of the effects of automated guidelines based on clinical data