PK/LO Data Warehouse
Home > Key Projects
The PK Warehouse project is aimed at building a single enterprise wide repository for all post-EDC compounds in a regulatory compliant (21CFR11). Environment. The goal of this project is to consolidate and integrate Time-Concentration, Pharmaco-Kinetic, Pharmaco-Dynamic, Descriptive Statistics and other Statistical Analysis results from various pre-clinical and clinical data sources, generated either internally or externally by CROs. INTEK was contracted to deliver in all phases of the project including gathering detailed requirements, developing specifications, evaluating and recommending the solution architecture, build and deploying a production pilot that includes capabilities listed below.
- Ability to transfer data from foundation data sources, present for user review and load into the warehouse.
- For Demography data and Actual Dose and Actual Sampling Time the transfer of data was required to be automated.
- The data from external CROs were required to be processed by the same business rules as data from internal sources
- Ability to register STUDY protocol data necessary to interpret the results
- Ability to integrate with WATSON, ClinTrial, Debra, WINNONLIN, SAS and CROs.
- Ability to integrate with WINNONLIN and generate input for NONMEM
- Ability to store associated documents and meta-data for all data being processed
- Ability to graph on-the-fly selected data that is being presented to the user and save this as a vector/JPEG image
- Ability to prepare and submit ad hoc queries to the Warehouse
- Ability to run scheduled reports and on-demand reports
- The system was built with out-of-the-box tools for the most part with customization limited to customer specific workflows and integration.
Predictive Foundation System (PFS)
The goal of this project was to automate the computation and storage of In-Silico properties on a nightly basis as and when new compounds are added to the chemical repository. INTEK was contracted to handle to design and build the data warehouse and associated ETL for processing and capturing the results.
- The entire system was custom built in Java, XML and Oracle 8I
- The system uses very advanced techniques for automatic data auditing and error detection
- System was designed to be very reliable and fault tolerant and can scale to large volume via multi-threading and multi-pass job-processing
- System was specified (including requirements gathering), designed, built and put in production on an aggressive time scale of 6 months
Data Warehouse for Activity Base (ABASE)
This collaborative project, with the vendor who developed ABASE, involves designing a new Star data model for a biological data warehouse, using Oracle 8i , and prototyping a global implementation. Key features of the project include separating the 'Operational' database from the 'Retrieval' database, developing new summary tables and prototyping the ETL process. In addition, limited data scrubbing was also involved. A major secondary goal was dissecting the performance contribution from various data modeling and database design criteria through a series of well planned, systematic studies. Key responsibilities include
- Project planning and initiation
- Requirements gathering and analysis; Criteria for separation, migration, data scrubbing and summarization
- Data modeling; Cardinality analysis, indexing recommendation and partitioning; Used Erwin 3.51
- Performance and Query tuning
- Design of materialized views (New feature of Oracle 8i)
- Designing the systematic performance evaluation studies
- Database consolidation and scaling to measure performance degradation
- Systematic performance evaluation of the new model