Background:
Cardiopulmonary bypass (CPB) is a cornerstone of modern cardiac surgery, requiring continuous, high stakes decision making based on rapidly evolving physiological and technical parameters. Despite advances in perfusion technology, CPB management remains highly dependent on the expertise of the perfusionist and is associated with significant complications, including acute kidney injury, myocardial injury, impaired oxygen delivery, and metabolic derangements. Recent progress in artificial intelligence (AI) and machine learning (ML) has introduced new opportunities to augment CPB management through data-driven prediction, monitoring, and decision support.

Objective:
This article reviews and synthesizes current evidence on the role of AI in CPB, with a specific focus on its potential to enhance safety, monitoring, and decision-making from a perfusionist centered perspective.

Methods:
Published studies employing machine learning and deep learning techniques in cardiac surgery, perfusion, and CPB related outcomes were reviewed. Emphasis was placed on models incorporating intraoperative and time series CPB data, external validation, and explainable AI frameworks relevant to perfusion practice.

Results:
Multiple Machine Learning and deep learning models have demonstrated superior performance compared with traditional statistical approaches in predicting CPB related complications, including acute kidney injury, perioperative myocardial injury, postoperative hyperlactatemia, and oxygen delivery deficits. Time series models leveraging minute level intraoperative data achieved the highest predictive accuracy, underscoring the importance of dynamic CPB variables. Explainable AI methods consistently identified clinically meaningful perfusion related predictors such as pump flow, oxygen delivery, lactate concentration, hematocrit, and cardiopulmonary bypass duration. Simulation based studies further suggest that ML algorithms can model perfusionist’s intraoperative decision making with acceptable accuracy, supporting the feasibility of AI assisted real time clinical decision support (Dias et al., 2022). Importantly, current evidence supports AI as an adjunct rather than a replacement for human perfusionists.

Conclusion:
AI has emerged as a reliable and clinically meaningful tool to support perfusionists during CPB by enhancing real time monitoring, predicting adverse outcomes, and assisting with complex decision making under human supervision(Dias et al., 2022). Given the structured, high frequency nature of CPB data, AI assisted perfusion represents a practical and incremental evolution in cardiac surgical care. Future research should prioritize prospective validation, integration into perfusion.