- Physicians and medical decision-support applications, such as for diagnosis, therapy, monitor- ing, quality assessment, and clinical research, reason about patients in terms of abstract, clini- cally meaningful concepts, typically over significant time periods. Clinical databases, however, store only raw, time-stamped data. Thus, there is a need to bridge this gap. We introduce the Temporal Abstraction Language (TAR) which enables specification of abstract relations involv- ing raw data and abstract concepts, and use it for defining typical medical abstraction patterns. For each pattern we further analyze finiteness properties of the answer set.