OR WAIT null SECS
Case Study: Turn data into knowledge.
Business Intelligence (BI) is the latest buzzword in healthcare, but in computer science terms, it is a combination of data mining and predictive modeling. Data mining is the unearthing of significant patterns and the ability to drilldown to find specific instances. Predictive modeling takes those patterns and extrapolates probable futures.
Managed care is very concerned with resource utilization and practice pattern variance correlated to outcomes and cost. A BI system can provide the tools to enable healthcare data analysts and program managers to dynamically explore this data, i.e., the ability to cross-tab practically any component of the episode against any other with unlimited selection criteria. This is a lot more than a set of standard reports that vary the date range.
BI consists of data mining and predictive modeling, which are built atop a data warehouse. If you didn't know much about computer science, one would expect such systems to be difficult, time-consuming, and expensive to design, construct, and implement. It doesn't have to be that way.
Many organizations will:
1. Hire a consulting firm to work with a project team in an understaffed IT department to understand data, design the data model, build the warehouse, and write custom reports.
2. Purchase licenses for the usual development software, i.e., Oracle or an equivalent database management package, SAS, Business Objects, Crystal Reports, etc., which the organization and the consulting firm will use to build a "unique" BI system.
3. Hope that they will be able to load data within six months and get some reports out of it about six months later.
The consulting fees, assorted licenses and tech support agreements will cost anywhere from a $250,000 to $1 million. Here are some observations that may help distinguish between myth and reality:
Myth: Healthcare data is especially difficult to store in a data warehouse. The fact is that healthcare data is extremely standardized, and its use of such classification structures as ICD9 and CPT codes makes it easier to model, not more difficult. The relationships between the various data elements are well understood. The predominant data source, from claims, has been standardized for years. Probably 90% of the data model for your "customized" warehouse is boilerplate.
Myth: Building a full-scale data warehouse all at once can be prohibitively expensive and time-consuming. Given the preceding comments about healthcare data, building a full-scale data warehouse all at once should be relatively quick. If the typical incremental, "data mart first" approach was taken, it is highly doubtful that your custom data warehouse will be able to integrate new data sources or adapt to the ever-changing healthcare landscape without substantial redesign. You've got to do the enterprisewide design first, not later. If it's well designed, then adding more reference tables, or incorporating eligibility or even EMR data, shouldn't be a challenge.
Reality:Advanced data analysis capabilities can improve care and cut costs. A 2,400 doctor IPA has been using our SmartCare software for four years to analyze claims data on a contract for about 50,000 covered lives. They've been able to increase their reimbursements every year. Each time they meet, the health plan's representatives say that their information is not as good as that presented by the IPA. The IPA gets all of its data from the health plan.
A Midwest consulting firm had gotten 1.2 million claim records from a large employer who wanted to improve its plan offerings. The firm decided to try Vantage Point's SmartCare data warehouse/analysis system, and contacted us on a Thursday. Sample data was provided to Vantage Point via secure server. The sample data was mapped and a conversion script written in three days. The script, along with a complete self-installing package for the system consisting of the data warehouse and the desktop analysis application, was put up on our secure FTP site, where the consulting firm downloaded it.