TY - JOUR
T1 - Machine learning for predicting mortality in transcatheter aortic valve implantation: An inter-center cross validation study
AU - Mamprin, Marco
AU - Lopes, Ricardo R.
AU - Zelis, Jo M.
AU - Tonino, Pim A. L.
AU - van Mourik, Martijn S.
AU - Vis, Marije M.
AU - Zinger, Svitlana
AU - de Mol, Bas A. J. M.
AU - de With, Peter H. N.
N1 - Funding Information: Funding: This research was funded for the Eindhoven university of technology and the Amsterdam university medical center by the European ITEA3 funding agency, project PARTNER (acronym of Patient-care Advancement with Responsive Technologies aNd Engagement togetheR), grant number 16017. Publisher Copyright: © 2021 by the authors.
PY - 2021
Y1 - 2021
N2 - Current prognostic risk scores for transcatheter aortic valve implantation (TAVI) do not benefit yet from modern machine learning techniques, which can improve risk stratification of oneyear mortality of patients before TAVI. Despite the advancement of machine learning in healthcare, data sharing regulations are very strict and typically prevent exchanging patient data, without the involvement of ethical committees. A very robust validation approach, including 1300 and 631 patients per center, was performed to validate a machine learning model of one center at the other external center with their data, in a mutual fashion. This was achieved without any data exchange but solely by exchanging the models and the data processing pipelines. A dedicated exchange protocol was designed to evaluate and quantify the model’s robustness on the data of the external center. Models developed with the larger dataset offered similar or higher prediction accuracy on the external validation. Logistic regression, random forest and CatBoost lead to areas under curve of the ROC of 0.65, 0.67 and 0.65 for the internal validation and of 0.62, 0.66, 0.68 for the external validation, respectively. We propose a scalable exchange protocol which can be further extended on other TAVI centers, but more generally to any other clinical scenario, that could benefit from this validation approach.
AB - Current prognostic risk scores for transcatheter aortic valve implantation (TAVI) do not benefit yet from modern machine learning techniques, which can improve risk stratification of oneyear mortality of patients before TAVI. Despite the advancement of machine learning in healthcare, data sharing regulations are very strict and typically prevent exchanging patient data, without the involvement of ethical committees. A very robust validation approach, including 1300 and 631 patients per center, was performed to validate a machine learning model of one center at the other external center with their data, in a mutual fashion. This was achieved without any data exchange but solely by exchanging the models and the data processing pipelines. A dedicated exchange protocol was designed to evaluate and quantify the model’s robustness on the data of the external center. Models developed with the larger dataset offered similar or higher prediction accuracy on the external validation. Logistic regression, random forest and CatBoost lead to areas under curve of the ROC of 0.65, 0.67 and 0.65 for the internal validation and of 0.62, 0.66, 0.68 for the external validation, respectively. We propose a scalable exchange protocol which can be further extended on other TAVI centers, but more generally to any other clinical scenario, that could benefit from this validation approach.
KW - Aortic valve disease
KW - Inter-center cross-validation
KW - Machine learning
KW - One-year mortality prediction
KW - Outcome prediction
KW - Prognosis
KW - Tavi
KW - Transcatheter aortic valve implantation
UR - http://www.scopus.com/inward/record.url?scp=85108523666&partnerID=8YFLogxK
U2 - https://doi.org/10.3390/jcdd8060065
DO - https://doi.org/10.3390/jcdd8060065
M3 - Article
C2 - 34199892
SN - 2308-3425
VL - 8
JO - Journal of cardiovascular development and disease
JF - Journal of cardiovascular development and disease
IS - 6
M1 - 65
ER -