MedTAKMI-CDI: interactive knowledge discovery for clinical decision intelligence

摘要: This paper describes MedTAKMI-CDI, an online analytical processing system that enables the interactive discovery of knowledge for clinical decision intelligence (CDI). CDI supports making by providing in-depth analysis data from multiple sources. We discuss fundamental challenges we faced and explain how met those developed a prototype experimental currently handles information about 7,000 patients at National Cancer Center in Japan. elaborate on three-layer model (attribute-value pairs, ordered sequences events, time-stamped events) information, which can represent three different levels abstraction. flexibility broad range analysis, simple demographic to mission-critical clinical-path pattern analysis. Rather than collection rigid relational schema our database employs metaschema with patient identifier, time stamp, attribute name, values. allows us modify representation without having reload rewrite analytic components. also describe functions are used understand care practice hospital, obtain overview navigate using layers abstraction ontologies, extract patterns rules paths.