CHICAGO – Dr. Eric Bremer knows a thing or two about brain tumors in children. The mortality rate is high, and the key to the most effective treatment is in knowing exactly what type of tumor is being treated.

Bremer, director of brain tumor research at Children's Memorial Hospital in Chicago, is using data mining and predictive analytics technology to help him more accurately classify the tumors.

An accurate classification of the tumor and its stage means treatment can be more exact. Giving no more and no less chemotherapy than necessary results not only in less pain, but it lessens the chance that the treatment will cause more damage.

"It's important not to over-treat," Bremer said. The challenge, he said, is to continually make sense of the 7,000 to 30,000 data points for each brain tumor sample.

Today, Bremer is building a gene expression database for pediatric brain tumors and matching it with research on effective treatments, and he's doing it faster than he could ever have done it using his previous method of manila folders and Excel spreadsheets.

The first stage of the work involved collecting all the data; the next hurdle was figuring out how to mine it.

Bremer found his answer in 2002 at a microarray data analysis conference when he met executives from SPSS, a Chicago-based company whose technology is used in government, financial services, communications and education, as well as healthcare.

"They introduced me to data mining," Bremer said.

John Quinn, a principal and chief technology officer in Capgemini's National Health Care Consulting Practice, has observed increased interest in data mining from his clients in recent months. In a hospital setting, data tends to reside in a number of silos, he said, so the challenge is to make sure all the data is accessible; otherwise even the best data-mining technology cannot be effective.

Bremer's use of SPSS' data-mining software, which it calls Clementine, is fairly routine and unique at the same time, said Catherine DeSesa, SPSS systems engineer. He uses the technology's capabilities in a way "not uncommon to those whom I consider advanced data miners," DeSesa said.

It's a rather unique use, she said, noting that Bremer is working with many of the advanced features of the technology. For example, custom user files enable the creation of rules to force the extraction and organization of complex concepts.

Bremer also uses SPSS' LexiQuest Mine to dig into volumes of biomedical literature to find more clues that could dictate treatment.

Bremer plans to continue his automated search and analysis of medical journals. Using SPSS technology and a United Devices grid computing system to speed processing, he was able to move through 124,000 medical abstracts in one hour and 18 minutes, compared to the 24 hours it would have taken on a single computer.