فهرست نشریات علمی دسترسی آزاد ایران

Aortic valve stenosis (AS) is the most prevalent valvular heart disease in developed countries, posing substantial clinical and economic burdens. Since its introduction in 2002, transcatheter aortic valve implantation (TAVI) has revolutionized treatment, necessitating meticulous risk assessment. Predictive analytics, particularly machine learning (ML), holds promise for forecasting health outcomes based on historical and real-time data. However, no systematic review has yet evaluated the application of predictive analytics in AS patients undergoing TAVI. To address this gap, a systematic search of Scopus, Web of Science, and PubMed/MEDLINE was conducted, identifying relevant studies published up to August 1, 2022. Seven studies met the inclusion criteria and were assessed using Qiao’s checklist for ML-based research quality. Findings were reported in accordance with PRISMA 2020 guidelines. The studies underscored the complexities of predicting post-TAVI mortality in AS patients, even with ML approaches. Some employed single ML techniques, while others utilized multiple algorithms, including artificial neural networks, Random Forest, Gradient Boosting, logistic regression, and automated ML. Input variables ranged from echocardiographic measurements and laboratory results to clinical symptoms and medical history. While mortality prediction was the primary focus, some studies also explored outcomes such as aortic regurgitation, valve sizing, and dyspnea improvement. Validation strategies included train/test splits, k-fold cross-validation, and leave-one-out methods. However, most studies relied on retrospective, single-center data with limited sample sizes and lacked external validation. Despite these constraints, ML-based algorithms show promise for enhancing cardiologists’ decision-making, provided their limitations are adequately addressed.