Practical Issues in and Applications of Clinical Data Mining

Pharmaceutical and biotechnology companies accumulate large volumes of data throughout the drug development life cycle—from discovery through marketing. Many organizations consolidate these data assets in central repositories to facilitate access and compilation for New Drug Application (NDA) submissions and other efforts. These clinical data repositories represent a potential gold mine of knowledge that may be further exploited for competitive business advantage. Clinical data mining solutions empower companies to extract previously unrealized information about their molecular compounds, product portfolio, clinical studies, and customers from data repositories. These companies must determine how to apply clinical data mining tools to get to the most useful information and, once knowledge is extracted, how to apply it for best business advantage. This paper provides an overview of the clinical data mining process and a framework for implementing a clinical data mining solution. The process begins with the data mining goal or objective and a database to be mined, followed by the application of data mining strategies to generate one or more models, and concluding with deployment of model results to relevant business areas. The process is illustrated in this paper through a case study using pooled studies from Phase I and Phase II clinical trials. Data mining should be a team effort of quantitative analysts (eg, statisticians), domain experts (eg, clinicians, scientists, drug discoverers), business users (eg, marketing and regulatory department), and information technologists, and must be thoroughly supported by upper levels of management and throughout the organization.