IBM, Mayo clinic team for medical database project

IBM will provide the Mayo Clinic with database tools, permitting doctors at the famed hospital to compare notes with every other case the facility has on record, it announced today in a statement. Their goal is to make every doctor a research scientist.

"The Mayo Clinic is special. Because it has so much data it's like a public system," said Jeff Augen, IBM's life sciences strategy director. Medical databases of comparable size to the Rochester, Minnesota-based clinic exist only in government medical databases outside of the US, or in countries with socialised medical systems like Canada, the UK or some Scandinavian states, Augen said. While large, those databases aren't as medically comprehensive as the Mayo Clinic's, he said.

The clinic runs a battery of tests on everyone it treats, be it for a bug bite or the plague. With 6 million patient records and 9.5 million more patients seeking treatment there every year, Mayo's databases are the most comprehensive in the world, Augen said. Two-thirds of those records were made after 1990. And most of Mayo's patients consent to their records being used for research, he noted.

However, the clinic's massive databases aren't relational, despite the reasonably mature information technology infrastructure at the hospital, he said. Records can't be crossed-checked against others to see if an illness befalls only people in a particular geographic area, demographic group or genotype.

The project from the Armonk, New York-based computer company essentially places the records of every consenting patient at the Mayo Clinic into a research pool. IBM copies the clinic's medical records, minus patient-identifying information, onto IBM pSeries servers running DB2 database software. The privacy-sanitised database can then be cross-checked using DB2 relational capabilities to correlate demographics, diagnoses and test results. This first phase is slated for completion by the end of June.

Augen hopes to extend the IBM database project into pharmacogenomic studies. With a large enough research population, the specific effect certain drugs have on patients with particular gene sequences can be determined by cross-checking treatment records with genetic records. Doctors could then match a therapy to a patient's genes.