Researchers have announced the start of a new project designed to identify Stanford patients who may have a genetic disease that causes a deadly buildup of cholesterol in their arteries.

Using big data and software that can learn to recognize patterns, researchers will comb through electronic medical records to identify patients at risk of familial hypercholesterolemia, which often goes undiagnosed until a heart attack strikes.

“This disorder certainly leads to premature death in thousands of Americans each year,” said Joshua Knowles, MD, PhD, assistant professor of cardiovascular medicine, who will lead the effort with Nigam Shah, PhD, assistant professor of biomedical informatics, and Ken Mahaffey, MD, professor of cardiovascular medicine. “Less than 10 percent of cases are diagnosed, leaving an estimated 600,000 to 1 million people undiagnosed. If found early enough and treated aggressively with statin-based regimens, people can live longer, healthier lives.”

The project is part of a larger initiative called FIND FH (Flag, Identify, Network, Deliver), a collaborative effort involving Stanford Medicine, Amgen Inc., and the nonprofit Familial Hypercholesterolemia Foundation to use innovative technologies to identify individuals with the disorder who are undiagnosed, untreated, or undertreated. The larger initiative is being funded by Amgen, a biotechnology firm that is developing an experimental cholesterol-lowering drug. The Stanford project is receiving additional funding from the American Heart Association.

Teaching a computer

For the Stanford project, researchers will use methods pioneered by Shah to “teach” a computer how to recognize a pattern in the electronic records of Stanford patients who have been diagnosed with FH. The computer will then be instructed to analyze Stanford patient records, for signs of the pattern. The researchers will then report their findings to the patients’ personal physicians, who can encourage screening and therapy.

“There are currently few systematic approaches to identify these patients,” said Knowles, who is the director of Stanford’s FH clinic. “The key is early identification in patients unaware of their condition before bad outcomes occur such as heart attacks.”

Machine learning, in which computer algorithms learn to recognize patterns within data, is widely used by Internet businesses such as Amazon and Netflix to improve customer experience, get information about trends, identify likes and dislikes and target advertisements, Knowles said.

“These techniques have not been widely applied in medicine, but we believe that they offer the potential to transform health care, particularly with the increased reliance on electronic health records,” he said.

“The general approach we’ll be pioneering has broad applicability in other arenas. We hope to make our algorithms applicable to several different electronic health record platforms, and the principles can be applied to other conditions.” If successful at Stanford, the project will be tested at other academic medical centers.

The project is an example of Stanford Medicine’s Biomedical Data Science Initiative, which strives to make powerful transformations in human health and scientific discovery by fostering innovative collaborations among medical researchers, computer scientists, statisticians and physicians.

Stanford researchers also will participate in other aspects of the larger initiative, including an effort led by the Familial Hypercholesterolemia Foundation to screen large aggregated databases for the disorder on a national level. Other components of the initiative include patient education, family screening and a directory of health-care providers with experience in treating familial hypercholesterolemia.

“FIND FH is a forward-thinking and groundbreaking initiative to identify individuals who are at profound risk of early and aggressive heart disease because they have FH, through the use of innovative technologies,” said Katherine Wilemon, founder and president of the foundation.

Click here to view the original article.