A PREFIX B+ TREE RELATIONAL MODEL OF THE CLINICAL DATA BASE ENVIRONMENT INCLUDING SECONDARY STORAGE AND ACCESS COSTS

An investigation of the utilization of a data base management system within the clinical environment is presented. The time oriented data model supported by the CLINFO system is studied and compared to the more general relational data model. The study presents rationale supporting the relational model for use in the clinical environment and presents a formal decomposition process to transform the three-dimensional CLINFO data model into a form suitable for a relational data base environment. The logical structure of the Urolithiasis Clinical Data Management System is presented and compared to the clinical characteristics of earlier research. An analysis of the physical characteristics of such a relational data base implemented using a prefix B+-tree files structure is presented. This structure is one of the native file structures on the TANDEM-16 multiprocessor computer system, on which CDMS is implemented. A storage model is developed that can effectively predict secondary storage costs for a relation in a prefix B+-tree environment. The model was validated using the Urolithiasis data base. A model is developed to study a set of high-level queries to a relational data base. A classification for one- and two-variable queries is proposed and algorithms for their effective evaluation are presented. Finally a model used to predict secondary storage access costs for the evaluation of a large class of queries is presented and validated using the Urolithiasis data base.