Heart disease is the biggest and most costly health care challenge in the United States. Heart failure is the primary cause of more than 55,000 deaths per year, according to data cited by the Centers for Disease Control and Prevention. It costs the country an estimated $34.4 billion each year.
Now researchers are applying predictive analytics to design diagnostics and evaluate treatments. IBM announced Oct. 9 that the National Institutes of Health has awarded it and two health care organizations a $2 million grant to develop predictive analytics for primary care doctors to identify patients at risk of heart failure as much as two years ahead of time. IBM is working on this research project with Sutter Health, a network of doctors and hospitals in Northern California, and Geisinger Health Systems, a health care services organization in Pennsylvania.
On this episode of the Data Informed podcast, Shahram Ebadollahi, the program director of health informatics research at IBM, explains details about research project.
The goal is straightforward. By identifying people at risk of heart failure six months or two years ahead of time, doctors can introduce preventative treatments. They can also collect data about how these treatments work, to improve the quality of health care.
Ebadollahi says the joint effort with Sutter and Geisinger is an important aspect of the project, because it combines the work of researchers in the lab with experts in the field. He explains the opportunities of working with the health care organizations’ medical experts and the challenges presented by the unstructured data in electronic health records.
He also explains the ebbs and flows involved in this big data science project and how predictive analytics research into heart disease can contribute to advances in understanding other conditions like diabetes and attention deficit hyperactivity disorder.
While this interview occurred on October 11, 2013, during the federal government shutdown, an IBM spokeswoman said the research grant was awarded before the shutdown happened and does not affect this project.
This project fits into a broader IBM effort to bring its analytics to health care. Stephen Gold, vice president of IBM Watson Solutions, noted in a recent podcast that natural language processing and machine learning systems are suited to digesting and analyzing large repositories of unstructured data. The company also announced on Oct. 11 that UNC Health Care was using its systems to manage medical data as part of a program to reduce the number of patients who are readmitted to the hospital.
Michael Goldberg is the editor of Data Informed. Email him at Michael.Goldberg@wispubs.com.
Stethoscope image from Wikipedia user Huji. Used under Creative Commons license.
Correction, October 15, 2013: The original headline of this article referred to predicting heart attacks. The research project is focused on predicting heart failures.