Medical data mining: The search for knowledge in workers' compensation claims

Workers' compensation claims data are an underutilized source of research data. Data mining is a step in the process of knowledge discovery. This report is an inquiry into the question of whether data mining can be used to expand the research potential of workplace injury claims. A descriptive data-mining method is used to avoid the hazards of predictive data mining. A key decision in descriptive data mining is the selection of the descriptive variables to be included. A major purpose of claims processing is documentation of benefit decisions. This documentation is used in selecting the case records embedded in a claim history. This novel process is called beneficiation not only because of the obvious reference to benefit but also because it is a mining term that describes a process that further refines metallic ores. Claim beneficiation identified four data forms that contain descriptive information about workplace injury claim data. They are the first report of injury, the activity prescription form, a functional capacity evaluation, and a physical capability estimate. When the data variables from these forms are combined they become a workplace injury case registry history which is a new frontier for research in injury epidemiology.