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

Abstract

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

Identifying Patients without a Survival Benefit following Transfemoral and Transapical Transcatheter Aortic Valve Replacement

“…This finding is consistent with evidence from the recent literature [PMID: 31000012, PMID: 33289422] and emphasizes the importance of correctly interpreting long-term myocardial sequelae rather than assessing LVEF alone in preinterventional risk assessment. As the overall incidence paradoxical low-flow–low-gradient aortic stenosis was rather low (<1% in the entire cohort), our finding supports the hypothesis that patients with a low LVEF and consecutively low pressure gradients across the aortic valve display worse postinterventional outcome after TAVR than patients with a low LVEF that can nevertheless generate high gradients across the aortic valve [ 28 , 29 , 30 ]. The absence of the binary variable low-flow–low-gradient aortic stenosis in our cohort as a significant predictor of futile treatment at 1 year indicates that pressure gradients are likely to have a higher sensitivity due to the relatively high cut-off value of 50% for LVEF in the current guidelines.…”

Machine Learning Prediction Models for Mitral Valve Repairability and Mitral Regurgitation Recurrence in Patients Undergoing Surgical Mitral Valve Repair

“…ML classifiers, in contrast to regression-based classifiers, account for unexpected predictor variables and associations between features and outcomes, facilitating recognition of predictors not yet described in the literature [ 44 ]. However, ML models are more data dependent than conventional statistical techniques, requiring a larger sample size for a modelling technique to generate a classifier with a good discriminatory ability.…”

A deep-learning approach for myocardial fibrosis detection in early contrast-enhanced cardiac CT images

“…To evaluate the classification performance, based on the ground truth evaluation built on the late CE-CCT images, a 5-fold cross-validation strategy ( 20 ) was performed (patient-wise) to reduce bias. For each of the 5 iterations in the validation, one-fold was used as the test dataset and the remaining four folds were used as the training dataset.…”