Dr. Matthew Samore, professor of internal medicine at the U. and chief of the Division of Clinical Epidemiology, is heading up Utah's part in the initiative, called the Consortium for Healthcare Informatics Research. Research teams of collaborating investigators are also located in Portland, Ore., Palo Alto, Calif., Indianapolis, Ind., Nashville, Tenn., Tampa, Fla., West Haven, Conn., Boston and Pittsburgh.
The initiative looks to use state-of-the-art language processing techniques to extract information from the health records that has been previously unavailable in easily searchable form.
It will allow researchers to "de-identify" the information so individual health information is not compromised, and pool it in ways that will help doctors understand better which treatments regimens are most effective in the greatest number of patients.
"It used to be that if you wanted to know something like how many patients in your hospital were treated with penicillin, you couldn't find out. Those electronic records were designed to take care of one patient at a time," rather than allowing researchers to look at pools of information from patients throughout the hospital or around the country, Samore said.
Medical informatics has changed that in the past 20 years, creating "clinical data warehouses that have been transformational," in furthering medical research, he said.
The consortium will take that information warehousing function to a new level, he said, providing rich detail including outpatient pharmacy records; lab reports; notes made by nurses, physicians, physical therapists and pharmacists; radiology reports and electromagnetic images; family history; discharge summaries; physician orders; vital sign measurements; and other factors that were previously inaccessible.
The $10 million, four-year initiative is an attempt to make the VA medical system records all "talk" to each other in deeper ways than they ever have before, Samore said, because at this point, "the (available) data tends to be stored locally. It's never been available at the national level," with the detail the initiative is designed to provide.
Individual VA hospitals have been trying to mine such information for some time, and have periodically worked with one other hospital doing the same, he said.
While major health care companies have long been at work on proprietary medical informatics software, the VA initiative has "adopted principles that favor, where possible, using open source software," Samore said. "That means these different components are organized in such a way that they are modular and can be shared."
Once the system has been developed and tested, researchers anticipate they will be able to find "novel associations" for disease and treatment that never would have been apparent previously.
"It might help us better identify the family and hereditary basis of disease or learn more about which treatment is better," he said.
Veterans have a disposition toward some types of medical problems, including PTSD, and researchers will conduct two multi-year applied studies to address both methicillin-resistant staph (known as MRSA) and PTSD, Samore said.
Because the research, and the initiative itself, are designed to provide better care for vets, it's vital to solicit opinions and priorities from veterans groups about what their priorities are, he said, adding researchers will ask national veterans groups for input.
"The concept is participatory research, that it should provide direct benefits to those who participate and they should have some involvement," Samore said.