We are currently developing a database implementation to support the temporal semantics of medical data for the T-HELPER physician workstation, a data- management and advice system for protocol-based care of patients who have HIV disease. To understand the requirements of a temporal querying language for the database, we have analyzed the types of temporal predicates found in the clinical-trial protocols. We have extended the standard relational data model with new temporal representations and operations to support these querying requirements. We define a history as a relational table that uses real-world time to store both instant-based events and interval-based states. We also develop a set of temporal operations on events and states as the basis for creating a relational algebra for histories. We use the set of algebraic operators to execute an extended SQL that supports protocol data-management requirements.
