Follow Datanami:
April 11, 2014

Data Analytics Used To ID At-Risk Heart Patients

George Leopold

A pilot project using predictive modeling of data, some of it “unstructured,” in electronic medical records helped identify thousands of patients in a Virginia health network at risk for developing heart failure.

Working with IBM and medical software vendor Epic, Carilion Clinic, based in Roanoke, identified 8,500 at-risk heart patients. The partners said the pilot project could lead to early intervention, better care and health care cost savings.

The pilot project used IBM’s natural language processing technology to analyze unstructured data, including clinicians’ notes and discharge documents. This medical data is frequently overlooked as hospitals and clinics struggle with management of medical records.

The partners said that analyzing clinical notes in the context of electronics medical records improves patient care while fostering early detection of chronic ailments like heart disease and underlying conditions like hypertension and diabetes.

The Virginia pilot project applied data analytics and predictive modeling to identify at-risk heart patients with a claimed accuracy of 85 percent. Of the 8,500 at-risk patients identified, IBM claimed the model uncovered 3,500 patients who otherwise would have been missed using traditional diagnostic methods.

With heart failure afflicting more than 5 million American adults, the Center for Disease Control and Prevention estimates patients will on average not survive five years after the first diagnosis. Heart failure consumes a growing percentage of the nation’s medical resources for patients 65 and older. The bill for treating U.S. heart patients is estimated to be $32 billon a year.

Hence, there is an urgent need for earlier detection and treatment as the U.S. population ages and health care costs soar.

“We’ve learned that predictive analytics insights from both structured and unstructured data is imperative to meet our goal of improving patient care at lower costs,” said Steve Morgan, MD, chief medical information officer at the Carilion Clinic.

The 8,500 patients identified in the Virginia pilot program were expected to develop heart disease within one year. Among the predictors were: physiological data like systolic blood pressure; use of prescription drugs like alpha and beta blockers; obesity; and lifestyle factors such as occupation.

All would become candidates for early intervention and care management, the partners said.

Carilion Clinic invested in an enterprise data center built by IBM two years ago to store electronic medical records on about 1 million patients. The pilot project pulled and analyzed records on 350,000 patients.

IBM said its content analysis software allows doctor’s using Epic’s electronic medical records software to incorporate their notes into patient records. Once added, predictive modeling can be used to identify patients at risk for heart failure.

The partners said they are now looking for ways to apply predictive analytics and natural language processing to early detection of other chronic diseases.

Related Items:

Data Sharing Provides Rx for Cutting ER Costs

Healthcare Agencies Ill-Prepared for Big Data, Report Says

Fighting Sepsis with Real-Time Analytics