Related Experiment Videos
Artificial intelligence and machine learning in global cardiac surgery: A scoping review
Shubh K Patel1, Jiawen Deng1, Adham Elsherbini1
1Temerty Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada.
Insights
Artificial intelligence and machine learning are advancing global cardiac surgery, especially for postoperative outcomes and mortality prediction. However, research is concentrated in China, highlighting the need for broader LMIC (low- and middle-income countries) involvement.
Area of Science:
- Cardiovascular Surgery
- Medical Artificial Intelligence
- Global Health
Background:
- Cardiovascular diseases are a leading global cause of death, disproportionately affecting low- and middle-income countries (LMICs).
- Access to advanced cardiac surgery remains a significant challenge in LMICs and remote regions.
- Artificial intelligence (AI) and machine learning (ML) offer potential solutions to improve diagnostics, surgical planning, and patient care, democratizing access.
Background:
Cardiovascular diseases remain the leading cause of death globally, with 80% of deaths occurring in low- and middle-income countries. Despite advancements in cardiac surgery, access to care remains limited in low- and middle-income countries and remote settings in high-income countries. Artificial intelligence and machine learning have the potential to democratize cardiac surgery access by reducing dependence on geography, while improving diagnostics, surgical planning, and peri- and postoperative care. This scoping review explores the application of artificial intelligence/machine learning in global cardiac surgery, focusing on low- and middle-income countries.
Methods:
A scoping review was conducted using Ovid MEDLINE, Embase, Web of Science, and Cumulative Index to Nursing and Allied Health Literature databases for studies on artificial intelligence/machine learning in cardiac surgery in low- and middle-income countries. Articles were summarized to assess trends, model use, and focus areas.
Results:
A total of 83 studies were included. The median sample size was 1521 (interquartile range, 523-5477). Most studies focused on adult populations (84.3%). Valve surgery (19.3%) was the most common domain followed by coronary surgery (15.7%) and congenital surgery (13.3%). Postoperative outcomes were the most common focus (55.4%), followed by mortality prediction (25.3%). Geographically, 74.7% of studies were from China, with limited representation from individual other low- and middle-income countries. XGBoost (25.3%) and random forest (16.9%) were the most commonly used machine learning models. Research increased substantially after 2020, particularly in postoperative outcomes and mortality prediction.
Conclusions:
Artificial intelligence/machine learning applications in global cardiac surgery are expanding, particularly in postoperative outcomes and mortality prediction. However, significant geographical differences persist, with most studies originating from China. Future research should prioritize expanding artificial intelligence/machine learning applications in other low- and middle-income countries and underexplored areas.
