AI Algorithm Raises Heart Attack Diagnosis Accuracy in UK Study

A computer algorithm developed using artificial intelligence accurately distinguished real heart attacks from false alarms, according to an emergency room study that offers a way for overburdened doctors to focus on the sickest patients.

The algorithm was able to accurately rule out a heart attack in nearly all patients, according to the study published Thursday in Nature Medicine. The tool, tested in more than 10000 patients in six countries, was 99.6% accurate in ruling out patients who weren’t suffering from heart attacks, performed well across different ages and genders.

Rapidly developing AI technologies are gaining traction in health care, aiming to streamline processes from drug development to direct patient care. Pharmaceutical companies are investing heavily in such tools to discover new drugs at a faster pace, and some experts are predicting applications for programs like ChatGPT in medical education, patient management and other areas.

Heart attacks are usually diagnosed by measuring levels of the blood protein troponin that’s released when heart cells die. The measurement, however, doesn’t take into account that certain patients, notably women, often have lower levels of troponin even when a heart attack occurs. The AI algorithm, CoDE-ACS, combines protein levels with individualized information such as age, sex, electrocardiogram results and medical history, to determine the probability of a severe event.

Many conditions can cause acute chest pain, one of the hallmarks of heart disease, and diagnosis isn’t always straightforward, said Nicholas Mills, a cardiology professor at the University of Edinburgh who led the research, in a statement.

“Harnessing data and artificial intelligence to support clinical decisions has enormous potential to improve care for patients and efficiency in our busy emergency departments,” he said.

Quickly ruling out a heart attack could reduce hospital admissions, researchers said. Clinical trials in Scotland are underway to determine whether the algorithm can help ease pressure on crowded emergency departments.

A computer algorithm developed using artificial intelligence accurately distinguished real heart attacks from false alarms, according to an emergency room study that offers a way for overburdened doctors to focus on the sickest patients.

The algorithm was able to accurately rule out a heart attack in nearly all patients, according to the study published Thursday in Nature Medicine. The tool, tested in more than 10000 patients in six countries, was 99.6% accurate in ruling out patients who weren’t suffering from heart attacks, performed well across different ages and genders.

Rapidly developing AI technologies are gaining traction in health care, aiming to streamline processes from drug development to direct patient care. Pharmaceutical companies are investing heavily in such tools to discover new drugs at a faster pace, and some experts are predicting applications for programs like ChatGPT in medical education, patient management and other areas.

Heart attacks are usually diagnosed by measuring levels of the blood protein troponin that’s released when heart cells die. The measurement, however, doesn’t take into account that certain patients, notably women, often have lower levels of troponin even when a heart attack occurs. The AI algorithm, CoDE-ACS, combines protein levels with individualized information such as age, sex, electrocardiogram results and medical history, to determine the probability of a severe event.

Many conditions can cause acute chest pain, one of the hallmarks of heart disease, and diagnosis isn’t always straightforward, said Nicholas Mills, a cardiology professor at the University of Edinburgh who led the research, in a statement.

“Harnessing data and artificial intelligence to support clinical decisions has enormous potential to improve care for patients and efficiency in our busy emergency departments,” he said.

Quickly ruling out a heart attack could reduce hospital admissions, researchers said. Clinical trials in Scotland are underway to determine whether the algorithm can help ease pressure on crowded emergency departments.