Artificial intelligence (AI) decision support systems could prevent unnecessary cardiac imaging tests for patients suffering from stable chest pain.
The Artificial Intelligence (AI) for Clinical Cardiac Navigation (ARTICA) decision support system (DSS) could prevent unneeded diagnostic tests for patients suffering from stable chest pain, according to an abstract presented at the 2019 International Conference on Nuclear Cardiology and Cardiac CT, held May 12-14 in Lisbon, Portugal.
“We know that doctors overtest patients and ignore recommendations about when a test is justified about two-thirds of the time,” said the study's lead researchers Marco Mazzanti, MD, of the Royal Brompton Hospital in London, United Kingdom in a statement. “Our ‘super brain’ decision support system, called ARTICA, strictly follows ESC [European Society of Cardiology] guidelines and does not advise unnecessary examinations.”
Researchers conducted the study to analyze differences in cost that could result from using ARTICA instead of standard care for patients with stable chest pain.
The ARTICA system used machine learning to make decisions that adhere to recommended practice. Researchers input guidelines for patients and routinely collected medical data. A machine learning algorithm analyzed data repeatedly until it learned to determine whether or not a patient required a test and identified which tests were needed.
“As doctors we order a lot of tests, which cost a lot of money and waste time. ARTICA is like a second set of eyes to make sure we follow recommendations,” Mazzanti said.
Researchers evaluated 982 patients with stable chest pain at 3 hospitals over a period of 16 months. Cost analysis was based on the ARTICA registry as well as the financial database of participating sites. Participants had a mean age of 54 ± 8 years and 58% were male. A human cardiologist and computerized automated DSS were present at each site and were consulted or used during same-day visits. In the study, significant coronary artery disease was defined as greater than or equal to 50% coronary stenosis on computed tomography angiography (CTA), performed on 961 participants, or on invasive coronary angiography (ICA), performed on 21 participants.
Researchers then compared decisions on which tests to perform made by either ARTICA or cardiologists. ARTICA advised no further testing (NFT) in 658 patients, or 67%, while cardiologists decided that only 45 patients, or 4.6%, required NFT. Researchers also found significant differences in exercise testing and functional imaging groups between DSS and cardiologists’ approaches.
CTA or ICA scans showed that 639 patients, or 97%, determined to require NFT by ARTICA had no significant coronary artery disease, implying the decision was valid. AI DSS showed a reduced number of medical procedures from a hospital perspective. Researchers found that avoiding unnecessary tests could save an average of 61 minutes for staff and 121 for patients.
Mazzanti also mentioned that ARTICA recommended exercise testing or functional imaging for 23% of patients while cardiologists recommended it for only 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.”
“AI has the potential to save costs and staff time by identifying patients with chest pain who do not have significant coronary artery disease and therefore do not need expensive cardiac imaging,” Mazzanti said.
Mazzanti M, A Goda, A Pottle, H Gjergo, E Shirka, F Pugliese. Cost analysis of cardiac imaging using artificial intelligence in subjects with stable chest pain. Results from the ARTICA database. Presented at: International Conference on Nuclear Cardiology and Cardiac CT. https://www.escardio.org/The-ESC/Press-Office/Press-releases/artificial-intelligence-could-prevent-unneeded-tests-in-patients-with-stable-che. Accessed June 3, 2019.