A recent scoping review highlights the growing role of artificial intelligence (AI) in the field of open-heart surgery.
It offers surgeons promising tools, providing real-time risk assessments and supporting personalised decision-making.
The review assessed the ability of AI to predict surgical results, improve patient outcomes, and guide personalised treatment plans.
The study, conducted in accordance with PRISMA-ScR guidelines, analysed data from 64 research articles sourced from PubMed, Web of Science, IEEE, and Scopus.
Findings reveal that AI techniques, particularly logistic regression, random forest, and XGBoost, are being increasingly used to forecast postoperative complications, including mortality and acute kidney injury.
Notably, XGBoost demonstrated the highest performance in numerous studies.
Despite these advances, researchers point out that deep learning and hybrid models remain underexplored.
The review also identified several challenges hindering clinical adoption, including inconsistent model validation, limited prospective data, and a lack of diversity in patient populations.
Experts emphasise that future efforts should focus on prospective validation, explainable AI, and ensuring data equity to improve the reliability and applicability of these models.
The authors conclude: ‘This scoping review emphasises the transformative potential of artificial intelligence (AI) in advancing risk prediction and clinical decision-making for cardiac surgery outcomes. While machine learning (ML) algorithms, particularly XGBoost and random forest, demonstrate superior predictive accuracy compared to traditional statistical methods, their clinical adoption remains hindered by methodological limitations, including the predominance of retrospective designs and insufficient validation in diverse patient populations. The field’s reliance on interpretable ML models reflects a pragmatic balance between algorithmic transparency and performance, though the underutilisation of hybrid and deep learning approaches suggests untapped opportunities for innovation in complex surgical scenarios.
To translate AI’s promise into clinical practice, future research must prioritise prospectively validated frameworks, standardised reporting of calibration metrics (e.g., Brier score), and the integration of explainable AI techniques to demystify “black-box” models.
Addressing the ethical and practical challenges of dataset bias, algorithmic generalisability, and equitable representation of vulnerable populations (e.g., paediatric and elderly patients) is equally critical. Collaborative efforts between clinicians, data scientists, and policymakers will be essential to develop robust, patient-centred AI tools that align with real-world surgical workflows. By bridging technical innovation with clinical rigour, AI can redefine precision medicine in cardiac surgery, optimising outcomes while ensuring trust, equity, and actionable utility at the bedside.’
Overall, AI holds promise for surgeons, they say. While current limitations exist, ongoing research and methodological improvements could usher in a new era of smarter, more precise cardiac care.
