A genomic CDS takes shape at Intermountain
Breakthroughs in genetic research represent an exciting frontier for medicine. But in these early stages, the need for more clinical data is clear. And the obvious place to look for this data is in hospital EHRs.
To explore the potential from merging these disparate collections of data, Intermountain Healthcare and Intel set out to build a clinical decision support (CDS) application based on all of the information available to the hospital.
“Today, we can do alerts with clinical decision support in our EHR. And we collect family health records that show risk factors that show the history of their relatives. When did they die? What did they die from? And we can look at genomic data, to see if a specific genomic variant sets a risk factor by itself,” said Grant Wood, senior IT strategist with Intermountain Healthcare’s Clinical Genetics Institute. “What if we brought all three together to form a combined data set? Could we define new risk algorithms and put them in a decision support application?”
Wood, in a presentation at the HIMSS15 Mobile Health Knowledge Center, said that the project is built on the current workflow.
A screener takes a sample workflow that takes about 3 to 5 minutes. That information is captured in a screening repository. If the patient is identified as high risk, a more complete family medical history is taken and stored in a repository. The high risk patients are then scheduled to meet with a genetic counsellor, who takes the family history and validates it with a patient screening application.
The genetic counsellors in their interview are trying to determine whether there should be more testing of the patient, or if the patient’s family members need further testing. The next step is genetic testing to determine risk factors. With all of the results available for analysis, decisions can be made about the need for treatment.
In today’s workflow, the delivery of data at all phases is generally done with PDF’s and the records and stored in separate databases. In moving toward more usable data repositories, the Intermountain project changed to database-friendly formats for reporting at all phases.
“Now we have all of these data repositories and it gives us the opportunity to do data discovery, and to go back and put these in decision support applications,” Wood said. “And that is fed back into risk assessment applications.”
The result is the ermergence of tools that provide the clinicians information they can use for treatment.
Wood shared the stage with Prashant Shah, healthcare architect with Intel Health and Life Sciences, who said that the project chose breast cancer for the initial work because it has the most well established treatment guidelines and historical data. “We want to make sure the system we build is based on standards and that we build these applications in EHR-agnostic applications.” All work in the project conforms to HL7 standards.
Intermountain was able to tap into its repository of 20 million records based on 342,000 patients and used Intel applications to generate risk data models.