Analytics integration spots unseen risks

December 3, 2013 in Medical Technology

A project at Carolinas HealthCare System to integrate data analytics for predictive modeling, individualized patient care and population health across the enterprise has seen encouraging early returns.

Carolinas HealthCare System comprises a hugely diverse network of providers – academic medical centers, hospitals, freestanding emergency departments, physician practices, surgical and rehabilitation centers, home health agencies, nursing homes, behavioral health centers, and hospice and palliative care services – across both North and South Carolina.

Recognizing that quality care is dependent upon the ability to deploy technology to understand data in a meaningful way, the health system created a comprehensive enterprise data warehouse to store 10 terabytes of patient data.

That’s helped with analytics advancements such as readmission risk modeling that can predict a patient’s 30-day readmission risk with nearly 80 percent statistical accuracy, officials say.

To get there, some 40 different patient variables that are deemed highly predictive of unplanned readmission are pulled each day from electronic medical records and analyzed; they’re then delivered in real time to healthcare providers who can prioritize high-risk patients and customize care for patients.

“At Carolinas HealthCare System, we have more than 10.5 million patient encounters in a variety of care settings each year and have used de-identified, rich data to develop innovative models that can be scaled broadly,” said Allen Naidoo, vice president of advanced analytics for Carolinas HealthCare System, in a press statement.

Outpatient analytics tools have enabled the health system to benchmark and monitor patient improvements across its facilities. Using EMR data to analyze care patterns and disease trends for more than 1.5 million patients, healthcare providers can segment and stratify, patient data by demographics, geography and clinical diagnoses.

This so-called geospotting analyzes the layers within a community and combines geographic information systems and predictive models to identify health needs and outcomes.

“Each data set we collect is masked of the original data contributor, but being able to drill down into the data helps us reduce variations in care and deliver overall better care, outreach and coordination,” said Michael Dulin, MD, chief clinical officer for analytics and outcomes research for Carolinas HealthCare System, in a statement.

Carolinas HealthCare System began incorporating clinical and financial data across its continuum of care in 2011 and later created an in-house advanced analytics group, Dickson Advanced Analytics, which is dedicated to creating strategic, long-term programs for developing and implementing healthcare analytics.

Earlier this year, the system also joined the Data Alliance Collaborative, an initiative aimed at improving population health on a national scale through data analytics and shared business intelligence.

The collaborative’s other members include Catholic Health Partners in Cincinnati, Fairview Health Services in Minneapolis, Texas Health Resources in Arlington, the Premier healthcare alliance and IBM work together to co-develop tools that integrate data from nearly 100 hospitals and more than 1,600 non-acute sites caring for 28 million people.

“In the end, healthcare analytics is about improving the health of the patient and of the population,” said Carolinas HealthCare System CEO Michael Tarwater in a statement. “To better understand and address their needs, we have to be able to see the bigger picture, and as we continue growing our capabilities, we can help patients more quickly and efficiently.”

[See also: Analytics means we 'roll up our sleeves']

Be the first to like.
VN:F [1.9.22_1171]
Rating: 0.0/5 (0 votes cast)

Article source:

Be Sociable, Share!
Bookmark and Share

Leave a reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>