Computers to mine Social Security data for work program

CHAMPAIGN, Ill. -- The National Center for Supercomputing Applications will spend a year analyzing Social Security Administration data to create a computer model aimed at helping disabled people return to the work force.

The University of Illinois' Disability Research Institute is funneling about $930,000 to the NCSA to examine the Ticket to Work program, an initiative designed to reduce the number of people who are dependent upon the Social Security Disability Insurance and Supplemental Security Income programs.

Ticket to Work offers SSDI and SSI beneficiaries "tickets" that they can redeem for employment services from various agencies, educational institutions and employers with on-the-job training. The service providers are paid based on the long-term employment success of their clients.

The problem is few people participate, said Thomas Prudhomme, director of NCSA's Cybercommunities Division.

He and Michael Welge, head of NCSA's Automated Learning Group, plan to analyze data from the first 13 states in which Ticket to Work was rolled out.

They'll study who received the tickets, whether they used them and what services they received. They hope to use the data to create a computer model that predicts which of the more than 10 million people receiving disability assistance are most likely to benefit and how to get more people participating.

The project is a new kind of venture for both the supercomputing center and Social Security.

The UI-based NCSA is known more for manipulating scientific data on things like the cosmos, the global climate and cell metabolism than the social sciences and societal issues. Most studies by the Social Security Administration tend to be based on results from surveys.

Prudhomme and Welge will use Social Security's own data to look not just at what people say but what they do.

"Our approach is to look at the data that Social Security actually has about their operations," Prudhomme said. If this project succeeds like we expect it to, Social Security is ready with open arms to do lots of things like this."

Key to the study will be the NCSA's Data-to-Knowledge, or D2K, software, which can be used to examine data in many different ways.

The software "can be applied to practically any domain where they're collecting data," Welge said. "Social Security has a lot of data. They just don't have the tools to manage and analyze it."

On the Net

National Center for Supercomputing Applications: www.ncsa.uiuc.edu