Mercy’s Big Data Project Aims to Boost Operations Making the most of the data it collects is a challenge for any organization, and those in the healthcare industry are no exception. Based in St. Louis, Missouri, Mercy health system includes 46 acute care and specialty hospitals, with more than 700 outpatient facilities and physician practices in Arkansas, Kansas, Missouri, and Oklahoma. With more than 40,000 employees, including over 2,000 physicians, Mercy’s vision is to deliver a “transformative health experience” through a new model of care. With such ambitious goals, Mercy has a compelling interest in harnessing the power of the data it collects. To do so, the health system needed to overhaul is data-management infrastructure and move into the world of big data. To make that move, Mercy partnered with software provider Horton works to create the Mercy Data Library, a Hadoop-based data lake that contains batch data as well as real-time data (stored in H Base, a distributed non relational database structure) from sources such as the Mercy’s ERP and electronic health record (EHR) systems. According to Paul Boal, director of data engineering and analytics at Mercy, “The blending of base batch data and real-time updates happens on demand when a query is run against the system.” Mercy’s new Hadoop environment, which contains information on more than 8 million patients, holds over 40 terabytes of data housed on 41 servers spread out over four clusters. Outside of improving patient care, a primary motive for the move to Hadoop was to improve Mercy’s administrative efficiency, particularly in the areas of medical documentation and claims generation. Ensuring that physicians, nurses, and lab staff complete the necessary documentation for a patient prior to discharge improves the chances that the hospital will generate an accurate and complete claim-reimbursement request. Prior to its Hadoop implementation, the health system had already initiated an automatic-documentation-review process. Now, Mercy plans to make use of real-time data along with the power of Hadoop to further improve upon this process. For instance, documentation specialists can generate reports that help them follow up with physicians regarding missing documentation during each morning’s clinical rounds. The hospital expects the new system will generate more than $1 million annually in new revenue based on claims that accurately reflect hospital patients’ diagnoses and treatment. Mercy is also focusing the power of its new technology on areas directly related to clinical care. “What we’re building out is a real-time clinical applications platform, so we’re looking for other opportunities to turn that into decision support,” says Boal. One such project involves leveraging the Hadoop environment to make better use of data generated by the electronic monitors in the intensive care units (ICUs) across the health system. Mercy now gathers 900 times more detailed data from its ICUs than it did before its implementation of Hadoop. The previous database system was only capable of pulling vital sign information for Mercy’s most critically ill patients every 15 minutes; the new system can do it once every second. The goal is to use the real-time data for better analysis, such as refining the health system’s predictive models on the early-warning signs of life threatening medical problems in the ICU setting. Like all healthcare providers, Mercy is required to maintain an audit trail for its EHR system. The audit trail keeps track of everyone who accesses any piece of patient information via the EHR. In addition to satisfying this regulatory requirement, Mercy expects Hadoop will help it put that audit trail data to a new use—analyzing staff behavior patterns and developing a better understanding of how processes actually get done. And in another Hadoop related project, lab staff are now able to quickly search through terabytes worth of lab notes that were previously inaccessible.
Critical Thinking Questions
1. One of the advantages of a Hadoop implementation is that it provides a high level of computing redundancy. Why is that particularly important in a healthcare setting?
2. Explain how the three characteristics of big data (volume, velocity, and variety) apply to the data being collected by healthcare providers such as Mercy.
3. How might Mercy benefit from an enterprise data model? Does Mercy’s move into big data make it more or less important that it have a clearly developed model that states the organizations’ data needs and priorities?