Adlassnig Klaus-Peter, Blacky Alexander, Koller Walter
Section on Medical Expert and Knowledge-Based Systems, Medical University of Vienna, Spitalgasse 23, A-1090 Vienna, Austria.
Stud Health Technol Inform. 2009;149:103-10.
Nosocomial or hospital-acquired infections (NIs) are a frequent complication in hospitalized patients. The growing availability of computerized patient records in hospitals permits automated identification and extended monitoring for signs of NIs. A fuzzy- and knowledge-based system to identify and monitor NIs at intensive care units (ICUs) according to the European Surveillance System HELICS (NI definitions derived from the Centers of Disease Control and Prevention (CDC) criteria) was developed and put into operation at the Vienna General Hospital. This system, named Moni, for monitoring of nosocomial infections contains medical knowledge packages (MKPs) to identify and monitor various infections of the bloodstream, pneumonia, urinary tract infections, and central venous catheter-associated infections. The MKPs consist of medical logic modules (MLMs) in Arden syntax, a medical knowledge representation scheme, whose definition is part of the HL7 standards. These MLM packages together with the Arden software are well suited to be incorporated in medical information systems such as hospital information or intensive-care patient data management systems, or in web-based applications. In terms of method, Moni contains an extended data-to-symbol conversion with several layers of abstraction, until the top level defining NIs according to HELICS is reached. All included medical concepts such as "normal", "increased", "decreased", or similar ones are formally modeled by fuzzy sets, and fuzzy logic is used to process the interpretations of the clinically observed and measured patient data through an inference network. The currently implemented cockpit surveillance connects 96 ICU beds with Moni and offers the hospital's infection control department a hitherto unparalleled NI infection survey.
医院获得性感染（NI）是住院患者常见的并发症。医院中计算机化患者记录的日益普及使得能够自动识别并对NI迹象进行扩展监测。根据欧洲监测系统HELICS（源自疾病控制与预防中心（CDC）标准的NI定义），开发了一种基于模糊和知识的系统，用于在重症监护病房（ICU）识别和监测NI，并在维也纳总医院投入使用。这个名为Moni（用于监测医院感染）的系统包含医学知识包（MKP），以识别和监测血流、肺炎、尿路感染及中心静脉导管相关感染等各种感染。MKP由采用Arden语法的医学逻辑模块（MLM）组成，Arden语法是一种医学知识表示方案，其定义是HL7标准的一部分。这些MLM包与Arden软件非常适合纳入医院信息或重症监护患者数据管理系统等医学信息系统，或基于网络的应用程序中。在方法方面，Moni包含扩展的数据到符号转换，具有多层抽象，直到达到根据HELICS定义NI的顶层。所有包含的医学概念，如“正常”“增加”“减少”或类似概念，都通过模糊集进行形式化建模，并且使用模糊逻辑通过推理网络处理对临床观察和测量的患者数据的解释。目前实施的驾驶舱监测将96张ICU病床与Moni连接起来，为医院感染控制部门提供了前所未有的NI感染调查。
