New! Sign up for our free email newsletter.
Science News
from research organizations

Artificial Intelligence Improves Heart Attack Diagnosis

Date:
September 16, 1997
Source:
American Heart Association
Summary:
Drawing on artificial intelligence technology, researchers have for the first time found that machines show promise of improving on human's ability to diagnose heart attacks, according to a study in today's American Heart Association journal Circulation.
Share:
FULL STORY

DALLAS, Sept. 16 -- Drawing on artificial intelligence technology, researchers have for the first time found that machines show promise of improving on human's ability to diagnose heart attacks, according to a study in today's American Heart Association journal Circulation.

Called "artificial neural networks," the computer-based method was more accurate than the cardiologist in reading the electrocardiogram (ECG), a test used to diagnose heart attacks in patients seen for chest pain in hospital emergency departments. The study was reported by Lars Edenbrandt, M.D., Ph.D., and co-author Bo Heden, M.D., Ph.D., of the University Hospital, Lund, Sweden.

"The neural networks performed higher than an experienced cardiologist, indicating that they may be useful as decision support," says Edenbrandt, a consultant in the department of clinical physiology at the University Hospital.

Neural networks are designed to "think" like humans, drawing knowledge and decision-making capabilities through experience. To teach a neural network how to recognize heart attacks, researchers exposed the computer memory to thousands of electrocardiogram readings, "more than any cardiologist could possibly read in a lifetime," notes Edenbrandt.

In the study researchers included 1,120 ECG records of people with heart attacks and 10,452 ECGs records that were normal. The neural networks were found to be 10 percent better at identifying abnormal ECGs than the most experienced cardiologists on staff.

An estimated 25 percent of ECG readings are "misjudged or overlooked" by the physician, and a person may be sent home from the hospital without a correct diagnosis, according to the scientists. However, the technology still won't replace a skilled physician who understands the fine points of the "the art of medicine because the ECG reading is only one of several tests used by physicians to diagnose a heart attack. Doctors will still need to talk to patients about their symptoms and medical history," he says.

Other co-authors are Hans Ohlin, M.D., Ph.D., and Ralf Rittner, M.Sc.

Circulation is one of five medical journals published by the Dallas-based American Heart Association.


Story Source:

Materials provided by American Heart Association. Note: Content may be edited for style and length.


Cite This Page:

American Heart Association. "Artificial Intelligence Improves Heart Attack Diagnosis." ScienceDaily. ScienceDaily, 16 September 1997. <www.sciencedaily.com/releases/1997/09/970916055603.htm>.
American Heart Association. (1997, September 16). Artificial Intelligence Improves Heart Attack Diagnosis. ScienceDaily. Retrieved March 18, 2024 from www.sciencedaily.com/releases/1997/09/970916055603.htm
American Heart Association. "Artificial Intelligence Improves Heart Attack Diagnosis." ScienceDaily. www.sciencedaily.com/releases/1997/09/970916055603.htm (accessed March 18, 2024).

Explore More

from ScienceDaily

RELATED STORIES