? Dell's blog needs full-time support |
| Online reviews? How much will you pay me? ?
September 03, 2006
The Mayo Clinic: An information company
I'm at the Mayo Clinic, in Rochester, Minn. I came here largely to learn about an alliance between Mayo and IBM to mine a century of medical records looking for patterns in disease, therapy and customer treatment. Mathematicians and computer scientists may find valuable patterns in that mass of data that lead to breakthroughs. (They'll be focusing on aggregate numbers, not on individuals.)
While here I learned that the very success of the Mayo Clinic owes a lot to innovation in information management. In 1907, Dr. Henry Plummer and his assistant Mabel Root established a comprehensive system of medical records. Each patient had one and they moved around with the patient so that each doctor could see all the treatments they'd had. What's more, they were color-coded and bordered, which made them searchable for medical researchers. This was cutting edge data warehousing. Plummer also had a system of pneumatic tubes installed to zip these records from doctor to doctor. So Mayo was also a networking pioneer.
The medical industry is now deluged with data, from genomics to CT scan images and millions and millions of patient records. Around the globe, researchers are working on ways to mine that data for new drugs, new therapies, not to mention industrial efficiencies. They have fearsome privacy issues to wrestle with. But as Mayo demonstrated a century ago, those who master medical information rise to the top.
The problem, as I'm sure the folks at Mayo informed you, is that there is no standard format for medical information indexing/coding. ICD-9CM, SNOMED, LOINC, NDC and NDF-RT, HCSPCS and CPT-4, UMLS... the list of "standards" that aren't go on and on.
Did they happen to mention how they are managing the issues of integrating information formats - using HL7 or the like for transformations?
Posted by: alwin at September 3, 2006 01:48 PM
They said that "structuring" the data is the biggest job. But at least it's all their own data, which makes it a bit easier.
Posted by: steve baker at September 3, 2006 08:04 PM