Predicting Long-Term Mortality in TAVI Patients Using Machine Learning Techniques

Background: Whereas transcatheter aortic valve implantation (TAVI) has become the gold standard for stenosis treatment in high-risk patients, it recently been extended to include intermediate risk patients. However, mortality rate at 5 years is still elevated. The aim of present study was develop a novel machine learning (ML) approach able identify best predictors 5-year after TAVI among several clinical and echocardiographic variables, which may improve long-term prognosis. Methods: We retrospectively enrolled 471 patients undergoing TAVI. More than 80 pre-TAVI variables were collected analyzed through different feature selection processes, allowed identification with highest predictive value mortality. Different ML models compared. Results: Multilayer perceptron resulted performance predicting TAVI, an area under curve, positive value, sensitivity 0.79, 0.73, 0.71, respectively. Conclusions: presented assessment factors Fourteen potential identified organic mitral regurgitation (myxomatous or calcific degeneration leaflets and/or annulus) showed impact on