Journal of Zhejiang University SCIENCE  B

Accepted manuscript available online (unedited version)


Machine learning models for predicting thirty-day mortality following TAVR: a national study from the NTCVR cohort


Author(s):  Qifeng ZHU1, 2, 3, 4*, Jin LU1, 2, 3*, Jiayuan LI1, 2, 3, Feiyu WU1, 2, 3, Danqing YU1, 5, Yihan PAN1, Qijing ZHOU6, Chongzhou ZHENG7, Daxin ZHOU8, 9, Wenzhi PAN8, 9, Xianbao LIU1, 2, 3, 4, Jian'an WANG1, 2, 3, 4

Affiliation(s):  1Department of Cardiology, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou 310009, China 2State Key Laboratory of Transvascular Implantation Devices, Hangzhou 310009, China 3Heart Regeneration and Repair Key Laboratory of Zhejiang Province, Hangzhou 310009, China 4Transvascular Implantation Devices Research Institute, Hangzhou 310053, China 5Department of Cardiology, Cixi Integrated Traditional Chinese and Western Medicine Medical and Health Group, Cixi 315300, China 6Department of Radiology, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou 310009, China 7Department of Structural Heart Disease, Cardiovascular Medicine Center, Affiliated Hospital of Guangdong Medical University, Zhanjiang 524001, China 8Department of Cardiology, Zhongshan Hospital, Fudan University, Shanghai 200032, China 9National Clinical Research Center for Interventional Medicine, Shanghai 200032, China

Corresponding email(s):  Jian'an WANG, email: wangjianan111@zju.edu.cn, Xianbao LIU, email:liuxb@zju.edu.cn

Key Words:  Machine Learning; Transcatheter aortic-valve replacement (TAVR); Mortality; Predictive model


Share this article to： More <<< Previous Paper|Next Paper >>>

Qifeng ZHU1,2,3,4*, Jin LU1,2,3*, Jiayuan LI1,2,3, Feiyu WU1,2,3, Danqing YU1,5, Yihan PAN1, Qijing ZHOU6, Chongzhou ZHENG7, Daxin ZHOU8,9, Wenzhi PAN8,9, Xianbao LIU1,2,3,4, Jian'an WANG1,2,3,4. Machine learning models for predicting thirty-day mortality following TAVR: a national study from the NTCVR cohort[J]. Journal of Zhejiang University Science B,in press.Frontiers of Information Technology & Electronic Engineering,in press.https://doi.org/10.1631/jzus.B2500804

@article{title="Machine learning models for predicting thirty-day mortality following TAVR: a national study from the NTCVR cohort",
author="Qifeng ZHU1,2,3,4*, Jin LU1,2,3*, Jiayuan LI1,2,3, Feiyu WU1,2,3, Danqing YU1,5, Yihan PAN1, Qijing ZHOU6, Chongzhou ZHENG7, Daxin ZHOU8,9, Wenzhi PAN8,9, Xianbao LIU1,2,3,4, Jian'an WANG1,2,3,4",
journal="Journal of Zhejiang University Science B",
year="in press",
publisher="Zhejiang University Press & Springer",
doi="https://doi.org/10.1631/jzus.B2500804"
}

%0 Journal Article
%T Machine learning models for predicting thirty-day mortality following TAVR: a national study from the NTCVR cohort
%A Qifeng ZHU1
%A 2
%A 3
%A 4*
%A Jin LU1
%A 2
%A 3*
%A Jiayuan LI1
%A 2
%A 3
%A Feiyu WU1
%A 2
%A 3
%A Danqing YU1
%A 5
%A Yihan PAN1
%A Qijing ZHOU6
%A Chongzhou ZHENG7
%A Daxin ZHOU8
%A 9
%A Wenzhi PAN8
%A 9
%A Xianbao LIU1
%A 2
%A 3
%A 4
%A Jian'an WANG1
%A 2
%A 3
%A 4
%J Journal of Zhejiang University SCIENCE B
%P
%@ 1673-1581
%D in press
%I Zhejiang University Press & Springer
doi="https://doi.org/10.1631/jzus.B2500804"

TY - JOUR
T1 - Machine learning models for predicting thirty-day mortality following TAVR: a national study from the NTCVR cohort
A1 - Qifeng ZHU1
A1 - 2
A1 - 3
A1 - 4*
A1 - Jin LU1
A1 - 2
A1 - 3*
A1 - Jiayuan LI1
A1 - 2
A1 - 3
A1 - Feiyu WU1
A1 - 2
A1 - 3
A1 - Danqing YU1
A1 - 5
A1 - Yihan PAN1
A1 - Qijing ZHOU6
A1 - Chongzhou ZHENG7
A1 - Daxin ZHOU8
A1 - 9
A1 - Wenzhi PAN8
A1 - 9
A1 - Xianbao LIU1
A1 - 2
A1 - 3
A1 - 4
A1 - Jian'an WANG1
A1 - 2
A1 - 3
A1 - 4
J0 - Journal of Zhejiang University Science B
SP -
EP -
%@ 1673-1581
Y1 - in press
PB - Zhejiang University Press & Springer
ER -
doi="https://doi.org/10.1631/jzus.B2500804"


Abstract: Background: Transcatheter aortic-valve replacement (TAVR) has emerged as the preferred treatment for patients with aortic-valve stenosis (AS). However, risk assessment tools tailored to this patient group remain insufficient. Aims: We aimed to develop a machine-learning-based model for predicting thirty-day mortality risk in TAVR patients, using a national cohort from China. Methods: This multicenter, registry-based study (NTCVR) included 10,799 patients undergoing TAVR at 147 Chinese tertiary hospitals (November 2011 - August 2024). Patients were split into three sets: training (60%), internal validation (20%), and external validation (20%, two separate provinces). Oversampling addressed class imbalance (30-day mortality rate: 2.9%). Extensive feature selection and model development employed 15 feature subsets and 15 ML learners, generating 1,125 candidate models. SHapley Additive exPlanations (SHAP) analysis was used to further assess the influence of selected predictors and machine-learning models. Results: The optimal model combined Double Input Symmetrical Relevance (DISR) feature selection with a Support Vector Machine (SVM) learner. The Area Under the ROC Curve (AUC) of the optimal machine learning model was significantly higher than that of the Society of Thoracic Surgeons (STS) score in both the internal validation cohort [0.74 (0.67, 0.81) vs. 0.60 (0.53, 0.67), p < 0.05] and the external validation cohort [0.69 (0.62, 0.76) vs. 0.57 (0.51, 0.63), p < 0.05]. SHapley Additive exPlanations (SHAP) analysis of the variables included in the optimal model highlighted the contributions to mortality prediction of baseline alanine aminotransferase, creatinine, NYHA class, need for circulatory support, and non-elective procedure status. Conclusions: A simple and effective machine-learning-based model was developed for predicting thirty-day mortality in TAVR patients, offering a valuable tool for risk stratification in China.

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

Open peer comments: Debate/Discuss/Question/Opinion

<1>

Please provide your name, email address and a comment





Full Text:  <22>

CLC number: 

On-line Access: 2026-05-11

Received: 2025-12-07

Revision Accepted: 2026-03-12

Crosschecked: 0000-00-00

Cited: 0

Clicked: 44

Citations:  Bibtex RefMan EndNote GB/T7714

Journal of Zhejiang University-SCIENCE, 38 Zheda Road, Hangzhou 310027, China
Tel: +86-571-87952783; E-mail: cjzhang@zju.edu.cn
Copyright © 2000 - 2026 Journal of Zhejiang University-SCIENCE