Medical Clustering Utilizing Claims Data
Operations Research Center,
Massachusetts Institute of Technology
MIT Medical Department
D2Hawkeye
Georgia Institute of Technology
Electrical Engineering & Computer Science, MIT
Large claims databases coupled with modern data mining methods have the potential to address important questions in healthcare. We explain how statistical clustering, in particular, can be used for health care cost predictions. We use claims data for close to 400,000 members over three years, to provide rigorously validated predictions of health care costs in the third year, based on medical and cost data from the first two years. Furthermore we illustrate through two examples, involving nonsteroidal anti-inflammatory agents on one hand and estrogen and antidepressants on the other, that our clustering algorithm can lead to discovery of medical knowledge.
