As businesses grow more dependent on data to deliver insights and drive performance, many business and IT leaders recognize an urgent need to modernize both their infrastructure and their enterprise software.
For Sutter Health, a not-for-profit organization that operates 24 hospitals and 75 other medical facilities in Northern California, handling more than 11 million outpatient visits annually, the challenges of managing data effectively became acutely apparent after years of organic growth and a series of mergers and acquisitions.
“We needed to integrate a variety of backend systems and data warehouse silos so we could introduce a centralized enterprise data management platform and disbursed data science function,” explains Souvik Das, principle data scientist and big data architect in the Enterprise Data Management team at Sutter Health.
In February 2014, the organization embarked on a modernization effort and began to examine its options. Strong data integration and standardization were critical requirements in selecting a new system, along with a more versatile approach. In October of the same year, Sutter Health went live with a new IT data infrastructure that incorporated SAP HANA. In April 2015, it added Cloudera’s Hadoop.
A resulting data lake has introduced a high level of flexibility and agility, particularly for providers and case managers who are accessing and viewing analytics results and data. Yet, it also allows the organization to pull from external data sources including—but not limited to—the U.S. Environmental Protection Agency (EPA), NOAA, FEMA, weather information, census data, Twitter feeds, and other socioeconomic and social sources of data.
“With HANA, we have the ability to implement changes quickly and easily,” Das says. “As standards, rules and needs evolve, we evolve with them.”
Gaining Visibility Into Efficiency Levels and Revenue Targets
The data platform has radically redefined the way the Sutter Health operates. For example, one of the organization’s core key performance indicators revolves around a daily census, sometimes referred to as a “bed count.” It provides visibility into efficiency levels and whether the organization is meeting revenue targets.
However, “Each hospital has its own way of tracking and computing numbers because services are counted differently at different facilities,” he says. As a result, a certain amount of data manipulation is necessary to deliver accurate results.
In the past, the process required manual intervention. Today, the organization is standardizing definitions of key metrics so that the process occurs seamlessly and automatically through the platform.
The benefits have been remarkable. Sutter Health now pulls previously unprocessed data results from analytics data marts in an hour or less, compared to the more than six weeks this process took in the past. The organization has achieved report processing speed improvements exceeding 40 times and reduced overall build efforts between 40 and 80 percent because there is no need for schema on writes, aggregates and indices.
Finally, analysts and business leaders can build asset health visualization models within a few hours. This has led to overall improvements in care, including more effective end-of-life patient quality improvement strategies.
The biggest initial challenge, Das says, was getting different groups across the organization to agree to common metrics and the same definitions for each metric. Likewise, different groups sometimes had different ideas about security and data governance, so they had to arrive at a consensus on a number of important issues.
After a few hiccups in integrating HANA and Hadoop, the initiative has gone smoothly. In fact, Sutter Health is now looking at expanding the platform to include more robust analytics, reporting and mobile apps. “It is helping us drive our strategy of personalized medicine and evidence-based care forward,” Das concludes.