Uninterrupted Healthcare Insight Discovery
Electronic Medical Record (EMR) adoption, big data and other technology trends are generating large volumes and varieties of data for analysis. Data is not the limiting factor in transforming the business and practice of healthcare; managing the data is. If healthcare organizations want to use these technologies to create opportunities to differentiate, they must invest in the data itself.
When this doesn’t happen, data warehouse and analytics project participants may tackle one use case at a time. For example, they know that an administrator seeks to better understand admission trends so they need to know how many patients were admitted last week. As a separate use case, this team knows that care coordination teams are tasked with improving care for Congestive Heart Failure (CHF) patients, so they want to incorporate data into the warehouse that allows care coordination teams to look at CHF patients and learn how many CHF patients were admitted last week. As a third use case, the team building a data warehouse has a request for identification of referring providers so they incorporate referring provider as a field in the data model and data warehouse.
The problem with this approach is that analytic use cases are dynamic in nature. When a researcher or an administrator is neck deep in data, they start out by testing their hypothesis – this begins with a known question (or many) and these initial questions rapidly compound and evolve. This rapid evolution is what drives innovation, identifies new cures and – ultimately is most likely to transform healthcare. This type of insight relies on data — seamless, transparent and uninterrupted access to data. Insights don’t wait for IT to connect new data sources or for data quality to be vetted. At Informatica, we refer to this type of analytics as discovery analytics. Discovery analytics enables clinicians and business users to rapidly perform iterative hypothesis testing on integrated data from internal and external sources.
This means that users with proper security profiles can quickly move from asking these pre-defined insulated questions to asking complex integrated questions, such as “What was the total cost of care for Medicaid patients admitted last week in zip code 98107 with CHF who had also been admitted in the last 30 days? And, of these, who saw their primary care provider post discharge and also picked up their prescriptions?” This example is illustrated below:
To experience this type of success, healthcare organizations need to focus on data, not solely on applications such as the EMR. An application-centric (or EMR-centric) view of data provides an incomplete snapshot of the patient by limiting the view to data within each individual application. To improve patient outcomes and make differentiating decisions based on dynamic analytic use cases, organizations require a view of patient information across applications and locations, including (but not limited to) claims processing, registration, physician offices, labs and inpatient beds. Organizations can only achieve this data-centric view by releasing data from its application silos, investing in its connectedness and quality and making it available for self-service discovery analytics.
In a traditional analytics environment, the serial process of getting access to data prevents rapid analysis that responds to changing needs. Often, users have access to data either within the application itself or through a business intelligence tool. In between, the data resides in a vacuum where it is extracted, transformed and made available to analytics users based on requirements set forth in advance. However, if after accessing the data through an analytics tool of choice the analyst identifies a data quality issue, needs more data, or wants different data, they must go back to IT and must start the process from the beginning. This is slow, costly and frustrating to end users.
A better way to accelerate analysis is to connect healthcare applications with a defined industry data model that sits upon a proven data management platform. This approach complements existing applications and accelerates time to value and makes discovery analytics a reality.