IBM, Mayo Clinic Take Next Data Mining Step

Computerworld - IBM and the Mayo Clinic last week said they are moving into the second phase of a technology partnership aimed at using a database of patient records to foster what Dr. Nina Schwenk, who heads Mayo's IT operations, calls the Holy Grail of medicine: individualized patient treatment.
IBM and the Mayo Foundation for Medical Education and Research, the formal name for the Rochester, Minn.-based nonprofit health care organization, announced that they were teaming up in early 2002 . After initially putting the medical records of about 4.4 million patients treated at Mayo's three clinics into a DB2 database, IBM and Mayo have spent the past two years developing and testing a methodology for mining the data.
Now they want to analyze information about patients in new ways and compare individual medical records to data about other patients who have similar disease characteristics and genomic makeups, said Schwenk. IBM and Mayo hope to use the results to fine-tune treatments, she added.
For example, a doctor treating a patient for cancer could use the data mining system to discover the results of treatments given to the last 500 patients who had cancers that were located in the same spot and had identical genetic characteristics.
The individualized care made possible by such information could one day replace current blanket-treatment approaches based on standard medical protocols, according to Schwenk. Instead of treating cancer patients with a common stew of chemotherapy, drugs could be tailored to fit specific cases, she said.
IBM plans to use its Blue Gene supercomputer on the project to do advanced molecular modeling for researching diseases, said Drew Flaada, IBM's director of the partnership with Mayo. He added that Mayo researchers will use Blue Gene to do "deep computing and deep science," including mathematical modeling and a simulation of gene structures to help predict the behavior of diseases.
The Blue Gene system, which is located at an IBM facility in Rochester near Mayo's headquarters, has a theoretical computing capacity of 360 trillion floating-point operations per second. IBM and Mayo are doing development work on an IBM p650 Unix server that's owned by the clinic, and they use a higher-end p690 machine as the production system that stores the patient records in DB2. The servers are based on IBM's Power4 processors and run its AIX operating system.
Schwenk wouldn't disclose the investments being made by IBM and Mayo, but she said the amount of funding is "significant -- a lot of zeros on both sides."

The mayo clinic hopes to use data mining results to fine-tune medical treatments. Image Credit: Reuters