REVIEWER: Mark Ridinger, senior vice president and chief science officer for American Healthways.

BACKGROUND: American Healthways is the nation's largest, most experienced disease and care management company. American Healthways works with hospitals, health plans and physicians to improve patients' health, enhance the care experience and reduce the cost of care. We provide services to 465,000 health plan equivalent lives and the 50,000 physicians who care for them in all 50 states.

PLATFORMS: American Healthways runs SAS 8.2 and Enterprise Miner 4.1 on Windows 2000.

PROBLEM SOLVED: To provide better patient care and to reduce expenses, American Healthways wanted to augment its ability to identify high- risk patients in a way that facilitated the patient-doctor relationship. With SAS Enterprise Miner, American Healthways is able to use predictive modeling to identify patients who are trending toward a high-risk condition. By combining the data mining and model building of SAS with clinical information systems and patient interaction, American Healthways gains a seamless integrated information system that helps identify high-risk patients. The predictive models developed with SAS generate advanced risk scores for each patient, which can then be used by American Healthways' nurses to initiate patient relationships. Once the relationship is established, care managers intervene and establish a plan for the patients to manage their diseases.

PRODUCT FUNCTIONALITY: American Healthways uses Enterprise Miner to develop risk predictive models primarily from medical and pharmaceutical claims data, and demographic and health plan eligibility data. We have been very pleased at the speed with which we can build and assess multiple and competing models using this platform, and on very large data sets. We are looking forward to future updates of Enterprise Miner that may include genomic variable selection, multivariate adaptive regression splines and Java score code.

STRENGTHS: SAS Enterprise Miner's strength lies in its capacity to build multiple models and evaluate those modeling techniques side by side without programming expertise. Its ability and efficiency to access and visually demonstrate the impact of each model is of significant value. SAS is also faster, reducing overhead costs ­ what we get done in three days with SAS would have taken weeks before.

WEAKNESSES: The main weakness of SAS Enterprise Miner from our standpoint is the ROC curve function (receiver operating characteristic). One of our primary methods for model evaluation is comparing the areas under the ROC curve. With the help of a SAS consultant, we've developed a macro that calculates the area under the curve; however, having this process automated would greatly increase the efficiency of model. We would also like to see more advanced nonparametric variable selection routines included.

SELECTION CRITERIA: When you think of analytic tools for data mining, there are several products that come to mind. We chose SAS Enterprise Miner for its seamless integration into our current technology, its robust data mining capabilities and its flexibility for users with varied statistical backgrounds.

DELIVERABLES: We build models that identify health plan members that will have an acute event (e.g., hospitalization) or be likely to be high cost in a future time frame. SAS Enterprise Miner's toolbox contains all the analytical procedures necessary to complete this process. We rely on many of the reporting tools such as sensitivity lift charts and the MultiPlot node to help compare models and describe our data, respectively.

VENDOR SUPPORT: We asked for and received a full-day session in Nashville with Enterprise Miner course trainers from SAS prior to making a decision to buy Enterprise Miner. This was very helpful in terms of answering some very detailed questions from members of my team. Currently, we have several individuals who do an outstanding job of supporting us via telephone and e-mail. We initially had, and it would be nice to have again, a local go-to person. The best part of working with SAS is the help desk. They never leave you hanging, are very polite and exceptionally responsive.

DOCUMENTATION: The documentation contains an abundance of well-structured critical information that has facilitated the data mining processes at our company. A large portion of the learning process has relied on the documentation.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access