Author(s): Koprowski SP Jr, Barrett JS
Abstract Share this page
Abstract OBJECTIVES: We have created a data warehouse for human pharmacokinetic (PK) and pharmacodynamic (PD) data generated primarily within the Clinical PK Group of the Drug Metabolism and Pharmacokinetics (DM&PK) Department of DuPont Pharmaceuticals. METHODS: Data which enters an Oracle-based LIMS directly from chromatography systems or through files from contract research organizations are accessed via SAS/PH.Kinetics, GLP-compliant data analysis software residing on individual users' workstations. Upon completion of the final PK or PD analysis, data are pushed to a predefined location. Data analyzed/created with other software (i.e., WinNonlin, NONMEM, Adapt, etc.) are added to this file repository as well. The warehouse creates views to these data and accumulates metadata on all data sources defined in the warehouse. The warehouse is managed via the SAS/Warehouse Administrator product that defines the environment, creates summarized data structures, and schedules data refresh. RESULTS: The clinical PK/PD warehouse encompasses laboratory, biometric, PK and PD data streams. Detailed logical tables for each compound are created/updated as the clinical PK/PD data warehouse is populated. The data model defined to the warehouse is based on a star schema. Summarized data structures such as multidimensional data bases (MDDB), infomarts, and datamarts are created from detail tables. Data mining and querying of highly summarized data as well as drill-down to detail data is possible via the creation of exploitation tools which front-end the warehouse data. Based on periodic refreshing of the warehouse data, these applications are able to access the most current data available and do not require a manual interface to update/populate the data store. Prototype applications have been web-enabled to facilitate their usage to varied data customers across platform and location. The warehouse also contains automated mechanisms for the construction of study data listings and SAS transport files for eventual incorporation into an electronic submission. CONCLUSIONS: This environment permits the management of online analytical processing via a single administrator once the data model and warehouse configuration have been designed. The expansion of the current environment will eventually connect data from all phases of research and development ensuring the return on investment and hopefully efficiencies in data processing unforeseen with earlier legacy systems.
This article was published in Int J Clin Pharmacol Ther
and referenced in Journal of Computer Science & Systems Biology