Timing is everything: Temporal reasoning and temporal data maintenance in medicine

Both clinical management of patients and clinical research are essentially time-oriented endeavors. In this paper, I emphasize the crucial role of temporal-reasoning and temporal-maintenance tasks for modern medical information and decision support systems. Both tasks are important for management of clinical data, but the first is often approached mainly through artificial-intelligence methodologies, while the other is usually investigated by the database community. However, both tasks require careful consideration of common theoretical issues, such as the structure of time. In addition, common to both of these research areas are tasks such as temporal abstraction and management of variable temporal granularity. Finally, both tasks are highly relevant for applications such as patient monitoring, support to application of therapy guidelines, assessment of the quality of guideline application, and visualization and exploration of time-oriented biomedical data. I propose that integration of the two areas should be a major research and development goal. I demonstrate one integration approach by presenting a new architecture, a temporal mediator, which combines temporal reasoning and temporal maintenance, and integrates the management of clinical databases and medical knowledge bases. I present and discuss examples of using the temporal mediator for several of the application areas mentioned. I conclude by reemphasizing the importance of effective knowledge representation, knowledge reuse, and knowledge sharing methods to medical decision support systems in general, and to time-oriented systems in particular.