Moving time window aggregates over patient histories.

Abstract

Moving window concepts are used in temporal query languages for aggregate functions over the dimension 'time'. In the medical domain, aggregation of patient data over time windows builds a powerful mechanism within clinical database queries to satisfy a class of typical medical question formulations. Contrary to other fields, like the business domain for example, there is the additional need to synchronize time windows with the individual course of diseases rather than with the calendar system only. In this paper, we present several variants of shifting time windows over patient histories and suggest a set of essential options for a moving window clause. The proposed parameters for window creation as well as suitable default settings are discussed in the context of retrieving data from medical records.