The Heritage Provider Network, a managed care company that contracts with several medical groups in southern California, is launching a contest to develop the best algorithm for predicting which patients are most likely to be hospitalized within a year. The two-year competition begins in April, with a prize of $3 million. Contestants will be provided with a batch of de-identified claims to play with.
Despite the big pot, what is the likelihood of our top mathematical minds coming up with an answer? Insurance companies have been trying to crack this puzzle for many years but have been unable to do it. This is not like an X prize for the first manned private space flight or an electric car that can get 100 mpg. Predicting who will get sick enough to land in the hospital is harder than rocket science.
As a point of reference, the X Prize, health insurer WellPoint and the firm's charitable foundation offered $10 million in 2008 to anyone who could develop new ways of improving the quality of care and lowering its cost. The sponsors canceled the five-year competition about a year ago, allegedly because they were unclear how healthcare reform would affect the healthcare system.
To learn more:
- read this post from the Wall Street Journal's Health Blog
Emory begins $50M predictive medicine program
Dashboards bring predictive modeling to healthcare management