Applying Temporal Abstraction in Clinical Databases

Abstract

Temporal abstraction (TA) methods aim to extract more meaningful data from raw temporal data. The use of temporal abstraction is important for decision support applications in clinical domains which consume abstract concepts, while clinical databases usually contain primitive concepts. In this paper we propose a new approach for TA from temporal clinical databases: using inference graph, an extension of transition graph, as an implementation technique of a knowledge-based temporal abstraction system. We also describe a system, TDM, that integrates temporal data maintenance and temporal abstraction in a single architecture. TDM allows clinicians to use SQL-like temporal queries to retrieve both raw, time-oriented data and generated summaries of those data. The TDM system has been implemented and applied in monitoring the treatment of patients who have colorectal cancer.