182 FuzzyKBWEdit: Fuzzy Knowledge Acquisition for Weaning from Mechanical Ventilation in Intensive Care Units Christian Schuh 1 , Michael Hiesmayr 2 , and Klaus-Peter Adlassnig 1 1 Department of Medical Computer Sciences, Section on Medical Expert and Knowledge-Based Systems University of Vienna Medical School, Spitalgasse 23, A-1090 Vienna, Austria e-mail: {christian.schuh, kpa}@akh-wien.ac.at 2 Department of Cardiothoracic Anaesthesia and Intensive Care Medicine, University of Vienna Medical School Währinger Gürtel 18–20, A-1090 Vienna, Austria e-mail: michael.hiesmayr@akh-wien.ac.at 1. Introduction The process of weaning a patient with respiratory insufficiency from mechanical ventilation is complex and requires expertise obtained from long clinical practice. The adequate application of human expertise for computer-assisted weaning is a widespread problem in this setting. The knowledge acquisition component we developed is designed to formalize knowledge in an easier way. It helps the cooperating physicians to generate crisp and fuzzy knowledge bases for computer- assisted weaning in intensive care units (ICUs). 2. Objective Patients who require mechanical ventilation during surgery, at a time when they are deeply sedated, must be slowly weaned from mechanical ventilation after the operation to the point when they can breathe spontaneously. The adequate use of expertise for computer-assisted weaning is a common problem in this setting [1]. The knowledge acquisition component we developed is designed to formalize knowledge in an easier way. This knowledge acquisition tool, which is used for computer-assisted weaning in ICUs, is represented by a so-called fuzzy knowledge-based editor (FuzzyKBWEdit), that allows the cooperating physicians to carry out changes in the knowledge base directly at the ICU, without the assistance of a knowledge engineer (Figure 1). 3. Methods The crisp and fuzzy knowledge bases generated by the editor FuzzyKBWEdit consist of variables, values, and rules. The variables represent the physiological parameters and respirator settings. The values are described in terms of fuzzy sets and linguistic terms. The knowledge bases as well as various experimental versions are implemented as plug-in knowledge bases for the FuzzyKBWean frame program [2]. Our application focuses on building knowledge bases for patients who are