Researchers Use Machine Learning to Identify Genetic Disease Patients

In years gone by there was Where’s Waldo? Then there was Finding Nemo. Today, on a more serious note but important even to small children, there is FIND FH.

FH, or familial hypercholesterolemia, is a genetic predisposition to very high levels of LDL cholesterol (the bad cholesterol) that may first manifest itself as a fatal heart attack at a premature age. There are therapies, statin-based, that reliably improve outcomes for patients with FH. But there is a big problem: 90 out of 100 patients who have FH do not know they have it or perhaps even what it is. Many physicians are equally in the dark about the disease and therefore do not screen patients for it. FIND FH hopes to change all that.

In an ambitious tripartite campaign, an academic center (Stanford Medicine), a nonprofit advocacy group (the FH Foundation), and a pharmaceutical company (Amgen) seek to FIND (Flag, Identify, Network, Deliver) FH using computer-based algorithms to troll through large databases including electronic medical records and locate patients who likely have the disease. At Stanford, FIND FH is being led by Joshua W. Knowles, MD, PhD (assistant professor, cardiovascular medicine), Kenneth Mahaffey, MD (professor, cardiovascular medicine), and Nigam H. Shah, MBBS, PhD (assistant professor, biomedical informatics).

According to Knowles, “Our overall goal is to turn the tide on FH by increasing the diagnosis of this underappreciated, underdiagnosed, and undertreated condition, thereby preventing heart attacks and saving lives.”

FIND FH is financed by Amgen and by the American Heart Association through a National Innovative Research Award. The AHA describes the award’s objective as supporting “highly innovative, high-risk, high-reward research that could ultimately lead to critical discoveries or major advancements that will accelerate the field of cardiovascular and stroke research.”Read more

Source: medicine.stanford.edu; image: wired.com