The visual display of temporal information

The detection of temporal relationships among time-ordered patient data is an important, but difficult, clinical task. Large volumes of computer-stored clinical data offer the possibility of aiding in the early detection of subtle trends and states, but the presence of irrelevant data can obscure relevant findings and relationships. We present a formal system for representing complex temporal data as events on an abstract entity called a time line. We define five time line operations, SLICE, FILTER, OVERLAY, NEW, and ADD. For each operation, we precisely define the operator's effect on a time line, including exceptions and boundary conditions. In addition to our time line formalism, we describe an interactive environment designed specifically to help humans visualize temporal data. We have developed a database kernel and a graphical user interface that used our time line formalism and operations to support temporal manipulations. Using our formal system and our visualization environment, we describe two issues in the display and manipulation of temporal data: 1. the temporal granularity problem, and 2. the calendar mapping problem.