How AI could prevent unnecessary testing in patients with chest pain

An artificial intelligence “super brain” could help eliminate unnecessary diagnostic testing in patients who present with stable chest pain, according to a recent study, potentially saving physicians and patients significant time and money.

The work, led by Marco Mazzanti of the Royal Brompton Hospital in London and presented at the International Conference on Nuclear Cardiology and Cardiac CT May 12, tested the utility of a machine learning algorithm known as ARTICA (Artificial Intelligence for Clinical Cardiac Navigation) in cherry-picking diagnostic tests for patients concerned about chest pain. The condition is extremely common, resulting in thousands of panicked visits to the ER each year.

“We know that doctors overtest patients and ignore recommendations about when a test is justified about two-thirds of the time,” Mazzanti said in a release. “Our ‘super brain’ decision support system, ARTICA, strictly follows European Society of Cardiology guidelines and does not advise unnecessary examinations.”

ARTICA was trained on verified medical data and ESC recommendations for patients with stable chest pain. When applied to 982 such patients in Mazzanti et al.’s study, the algorithm advised no further testing in 67% of the group—a stark contrast to a cardiologist’s recommendation that just 4.6% of the same pool didn’t need further testing.

When patients referred for more tests by the cardiologist did undergo computed tomography angiography (CTA), those CT scans found 97% of the patients ARTICA said didn’t need more tests didn’t have any significant coronary artery disease, meaning its predictions were correct.

Mazzanti said avoiding extra imaging would save hospital staff one hour and patients two hours on average, as well as hundreds to thousands of dollars. A CTA exam, for example, could cost a patient up to $450 they didn’t need to spend.

Mazzanti said ARTICA also recommended exercise testing or functional imaging for 23% of patients, while cardiologists recommended it for just 10%.

“We know that when ARTICA says ‘don’t do a test’ it is almost 100% right because the CTA scan confirmed no blocked arteries,” he said. “When ARTICA decides a test is needed, we are less certain that this is correct. By adding more data to the super brain, these decisions will become more accurate and enable us to deliver more personalized care.”