Yuval Shahar

Querying and analyzing multiple time-oriented patient data is a key task during medical research, clinical trials or the assessment of the quality of therapy. In this paper, we present several aspects of the VISITORS system, which includes knowledge-based tools for graphical querying and exploration of multiple longitudinal patient records. We focus on the syntax and semantics of the knowledgebased aggregation query language for multiple time-oriented patient records, and on the graphical queryconstruction interface. The query language assumes an underlying computational method for deriving meaningful abstractions from single and multiple patient records, such as we had previously developed. The aggregation query language enables population querying using an expressive set of constraints. By using our underlying temporal mediator architecture, the time needed to answer typical temporal-abstraction aggregation queries on databases of 1000 to 10000 patients was reasonable.