"Predictive Modeling" Becomes Tool for Insurers

The Wall Street Journal September 12, 2002

In a way similar to how weathermen use computers to forecast coming storms, HMOs and other insurers are using computer technology to predict whether -- and when -- members are going to fall ill.

The programs use "predictive modeling" technology to crunch the multitude of personal and medical information that health insurers store about their members (eg, age, prescription drug use, gender, family history, blood test results, and doctors' reports). The resulting analysis can forecast how much an insurer will spend treating patients with diabetes or whether a new drug will keep patients with heart disease out of the hospital. It also can identify patients who are likely to have a medical complication, even if they appear completely healthy.

These companies insist that the new technology is not used to set insurance rates or terminate high-risk enrollees, but rather to pinpoint those who would benefit from preventive care. They say keeping people out of the hospital will save money for insurers and employers alike. Such major health players as Aetna, United Health Group, and CIGNA use predictive modeling. Another company embracing the concept is American Health Ways, a Nashville, TN disease management company that uses nurses and other medical personnel to keep tabs on patients with chronic conditions, advising them to watch their weight, check glucose levels, and so on.

Skeptics question the precision of the computer programs, and privacy problems abound. Critics fear the technology could be abused, especially by drug companies marketing preventive medicines and HMOs eager to protect themselves from people who are health risks.

Top - Back to Article Archive