Randomized controlled trials evaluating artificial intelligence in cardiovascular care - A Systematic Review

Introduction:
Artificial intelligence (AI) has shown promise in transforming healthcare, particularly in cardiology. However, high-quality evidence demonstrating its impact on critical clinical outcomes remains limited. This systematic review synthesizes evidence from randomized controlled trials (RCTs) evaluating the application of AI in cardiology, with a focus on key clinical outcomes.

Material and method:
We conducted a systematic review following PRISMA guidelines, searching MEDLINE, Web of Science, and the Cochrane Library from inception to October 2024. Eligible studies compared machine learning models with traditional methods in cardiovascular care. Primary outcomes included patient-important metrics such as mortality and hospitalization rates, while secondary outcomes focused on time and resource savings.

Results and discussion:
Eleven RCTs published between 2021 and 2024 were included, with 81.2% being multicenter trials. These studies investigated diverse AI applications in cardiology, including ECG (54.5%), computed tomography (27.3%), and echocardiography (18.2%). Five studies (45.5%) reported improved clinical outcomes such as reduced mortality, enhanced treatment adherence, and increased procedural success rates. Six studies (54.5%) demonstrated enhanced diagnostic accuracy and early detection of conditions like coronary artery disease, atrial fibrillation, and reduced ejection fraction, while three studies (27.3%) highlighted improved resource utilization, including reductions in unnecessary diagnostic tests, increased procedural efficiency, and clinician time savings. Subgroup analyses indicated variability in AI performance across different populations and study settings. However, potential bias related to funding sources was identified in 63.6% of studies, underscoring the need for more independent research to validate these findings.

Conclusions:
AI has the potential to enhance cardiovascular care by improving early detection, diagnostic accuracy, and operational efficiency. However, the limited number of high-quality RCTs highlights the need for larger, multicenter studies to validate these findings and ensure real-world applicability. Future research should emphasize robust trial designs, diverse populations, and standardized outcome measures to maximize AI’s clinical impact.