Humana Digs Deep Into Its Data With Predictive Modeling

Several years ago, Humana Inc. embarked on a multifaceted program to refocus its business more directly on the consumers of its health benefits plans—its corporate customers and their employees. The idea was to give them greater choice and control over their health, health benefits and budgets.

The $13 billion company had plenty of data on which to base its new ways of doing business: terabytes of information generated by traditional transaction-processing systems. Just one of the 432 tables in its enterprise data warehouse has 430 million rows of health claims data. But the systems that processed claims weren't well suited for the kinds of predictive and analytic tasks that Humana's underwriters, actuaries, marketers and medical specialists now faced.

What they required, it turned out, was the Insight Engine, a suite of four analytic tools developed by Humana: SmartStart is used by employers to evaluate alternate benefit plans and predict their financial results, VeriSight combines historical and predictive data to identify and understand health care trends, the Outcome tool provides insight into consumer behaviors, and the Health Case Management tool spotlights members with certain severe or chronic health problems so they can receive targeted help. In some cases the models can summarize millions of rows of data in just one color-coded display.

A Win-Win Situation

The models, which were completed last October, help Humana "identify markets, enhance products and predict costs," says Brian LeClaire, vice president and chief technology officer at the Louisville, Ky.-based company. "We connect data in ways we had not historically connected it. You are looking at thousands of pieces of information on a screen that, through colors, for instance, will draw your attention to specific patterns, disease categories, places of treatment and so on."

In insurance, the name of the game is setting rates as low as you can to be competitive, but not so low that you lose money. Carol McCall, vice president of Humana's Center for Health Metrics, says the company recently used VeriSight to set rates for a customer.

"The granularity of this information is much finer than they had had previously," she says. "Before, they would have gone out with a much higher rate increase than they ultimately did."

Because of high transaction volumes (millions of claims per month), plus the size of Humana's enterprise data warehouse (3.5TB), commercial extract, transform and load tools proved inadequate, leaving the company no choice but to develop its own. And the key visualization tool from Visual Numerics Inc. in San Ramon, Calif., proved inadequate off the shelf. But through close collaboration, Humana and the vendor were able to tune the product and redesign the application for high volumes, says Mitzi Silliman, an IT vice president.

"Humana, like many other organizations, is just starting to realize that information they have accumulated over the years ... can be used to positively impact patient care and also reduce medical error," says Scott Tiazkun, a health care analyst at IDC. "Technologies like predictive modeling and analytics are crucial to mine the vast amounts of information collected in medical records [and] can ultimately lead to more-appropriate medical treatments and reduce health care costs."

Humana Inc.

Business: Offers health insurance and related services to employers, government-sponsored plans and individuals.

Project champion: Brian LeClaire

IT department: 1,200

Project payback: A suite of analytic tools developed in-house enables Humana to understand its business like never before.

Special Report

Best In Class IT Leaders 2005

Stories in this report:

Copyright © 2005 IDG Communications, Inc.

It’s time to break the ChatGPT habit
Shop Tech Products at Amazon