The overall aim of this research is to develop an advanced data mining environment for extracting information relating to the safety of drugs (and other therapeutic products) from large databases of adverse reaction reports. The project builds upon prior research on a first-generation system which has received pilot use at FDA, CDC, and several pharmaceutical companies. This Fast Track proposal focuses on creating a next-generation system that achieves significant breakthroughs along several dimensions: computational and statistical methodology, ease-of-use and ease-of-access, data pre- and post-processing, and results visualization and interpretation. Such a next-generation system is an enabling technology for improved risk management, including early safety signaling and the investigation of drug interactions and adverse event syndromes. In Phase I, feasibility will be established through research in four key areas: (1) performance enhancement to handle larger, more complex problems; 2) results filtering to help with discriminating "interesting" from "uninteresting" results; 3) different baselines for use in comparison studies; and 4) specialized graphics for exploring and interpreting more than two-factor effects. Phase II will create a full-scale prototype of the proposed next-generation safety data mining system and will include several cycles of beta-testing and validation by collaborating statisticians and medical safety officers.
Thesaurus Terms: computer system design /evaluation, drug adverse effect, information retrieval, information system, statistics /biometry Internet, computer graphics /printing, computer program /software, drug interaction, mathematics clinical research
