34Cost analysis of cardiac imaging using artificial intelligence in subjects with stable chest pain. Results from the ARTICA database

Abstract

Background. Artificial intelligence (AI) applications like decision support systems (DSS) can prompt application of evidence based cardiology in clinical practice. Purpose. To perform cost analysis of DSS approach vs Standard Care (STD) in patients with stable chest pain (SCP). Methods. The cost analysis was based on the ARTICA registry and financial database of participating sites. Over a 16-month period, 982 patients (566 males / 416 females, age 54 ± 8 years) with SCP were evaluated in three hospitals. For the purpose of this evaluation, a computerized automated DSS and a human cardiologist at each site (STD) were applied during the same-day visit. Significant coronary artery disease (CAD) was defined as > =50% coronary stenosis on CTA (961 subjects) or on invasive coronary angiography (ICA) (21 subjects). Results. The DSS classified 658 (67%) patients as “No further test (NFT)”. In contrast, STD labelled as NFT only 45 (4.6%) patients. Significant differences were also found in Exercise testing / Functional Imaging groups between DSS and STD approach. Results are detailed in the Table. After CTA or ICA, 639 (97%) of NFTs identified by DSS showed no significant CAD. From a hospital perspective, AI DSS showed a reduced number of medical procedures. The average staff time was significantly reduced by a mean of 61 minutes (95% confidence interval (CI) 49 to 57) per patient, accompanied by a significant mean reduction of 121 minutes (95% CI 101 to 152) in the time patients stayed at the office. Conclusions. The AI DSS approach has the potential to save costs and staff time in the triage of stable chest pain patients. Due to an earlier identification of patients without significant CAD, the use of unnecessary cardiac imaging tests may be reduced. (Figure Presented).