Intermountain, Deloitte Tool Analyzes Health Outcomes
Big data software application OutcomesMiner can help researchers pinpoint specific shared characteristics of patients in various scenarios.
7 Big Data Solutions Try To Reshape Healthcare
(click image for larger view and for slideshow)
Intermountain Healthcare, a large provider organization based in Salt Lake City, and consulting firm Deloitte have launched a big-data analytics tool called OutcomesMiner. Drawing upon insights derived from Intermountain's clinical experience and large pool of electronic health record (EHR) data, the tool is designed to help healthcare providers, pharmaceutical firms and device makers understand how a wide range of factors might contribute to the outcomes of certain patients. It also can support comparative effectiveness and new product research, according to a news release.
"The life sciences and health care industry is entering a new era in which success is tied to demonstrating value for reimbursement with an increased focus on safety and clinical effectiveness," said Brett Davis, general manager of Deloitte Health Informatics, in the release. "OutcomesMiner can help our clients thrive under this new paradigm by providing the insights they need based on rigorous real-world evidence."
In an interview with InformationWeek Healthcare, Davis described Outcomes Miner as a tool that allows users to "understand clinical variations across multiple dimensions -- including different comorbidities and therapies -- and correlate those with outcomes in the population."
The researchers at Intermountain's Homer Warner Center for Informatics Research , which helped develop the tool, are "intrigued" by the results to date, said Katherina Holzhauser, assistant VP of IS commercialization for Intermountain, in an interview.
But she cautioned, "For us, it's a generator of hypotheses. It's not a direct clinical tool. It's for our researchers to look at and say, 'Here are the different signals in our clinical practice. Are any of them correlated to different outcomes between patient groups? And if there are correlations, are there other areas we should look at to see what is actually happening in that patient population?'"
What OutcomesMiner allows the researchers to do, she added, is to quickly find out which of their hypotheses are not supported by the data and what other possibilities might exist. "It's the first view into the power of big data to generate and validate or invalidate different hypotheses. So from that viewpoint, it definitely works."
Davis noted that the tool can be used as a fine data filter on particular patient cohorts. For example, he said, a typical benchmarking tool might show that an organization's diabetic patients are doing 15% worse than the national average without providing further insights. But OutcomesMiner can show that the rate of emergency room visits among these patients might be driven by those within a certain age range who also have hypertension and are on two particular drugs.
Intermountain already has held some discussions with pharmaceutical companies about OutcomesMiner, Holzhauser said, although she declined to name them. She said the conversations had been productive in that they "asked questions we might not have asked ourselves in the course of clinical care. So we can see where we may have gaps in data and begin collecting that information."
Holzhauser admitted that clinical analytics based on the Intermountain population might not reflect the population of other healthcare organizations. Intermountain would like to form a collaborative with other healthcare systems that use OutcomesMiner to compare insights. However, it will not pool its de-identified data with that of other organizations. "The goal is not to create a big database in the sky," Davis noted. "The goal is to create a healthcare collaborative of institutions that bring unique data and informatics to the table to create value."
Intermountain has had an EHR since the 1970s that contains about 90 million patient records, Holzhauser noted. Although for most purposes its researchers are looking at only 2.5 million records, collected over the past 12 years, that's still a very large database that exceeds the size of most healthcare organizations' repositories.
Asked how big an organization would have to be to use OutcomesMiner, Davis said that the statistical power of the tool depends not just on the quantity, but also on the quality of data. However, he said that clients do not necessarily need to apply OutcomesMiner directly to their own data to gain value from it.
"There are two ways that clients can take advantage of the OutcomesMiner platform," he said. "They can enable it onsite with their own data. They can also subscribe to a portal to use the same interface but with Intermountain insights and data, so they can look at their own population and data and compare it to Intermountain's."
Many healthcare organizations are starting to take advantage of the big data approach in various ways. For example, IBM just announced a collaboration with the Premier Healthcare Alliance and four healthcare systems to develop the analytic tools they'll need for population health management.