Predictive modeling is an emerging discipline that health plans use to identify members who are likely to become ill soon and are therefore likely to cost the plan a lot of money. The objective is to manage these patients prospectively to minimize expensive events and conditions. Predictive modeling isn’t trivial –it involves data scrubbing and sophisticated data mining.
But it turns out that making accurate predictions is only half the battle. According to Howard Brill, manager of medical informatics at Monroe Plan for Medical Care, Inc., as quoted by the Healthcare Intelligence Network:
One of the major challenges weâ€™ve had is communicating what that information means to our nurse case managers. There is a very strong pull by nurse case managers to focus attention on the sickest patients. These are not necessarily the individuals on whom they have the greatest impact.
It’s another example of how a systematic approach is necessary to achieve cost and quality improvements. Well intentioned individuals acting in a vacuum or sophisticated modeling techniques aren’t effective on their own.April 25, 2005