Machine-learning based prediction of in-hospital death for patients with takotsubo syndrome: the InterTAK-ML model

  • Ovidio De Filippo
  • Victoria L Cammann
  • Corrado Pancotti
  • Davide Di Vece
  • Angelo Silverio
  • Victor Schweiger
  • David Niederseer
  • Konrad A Szawan
  • Michael Würdinger
  • Iva Koleva
  • Veronica Dusi
  • Michele Bellino
  • Carmine Vecchione
  • Guido Parodi
  • Eduardo Bossone
  • Sebastiano Gili
  • Michael Neuhaus
  • Jennifer Franke
  • Benjamin Meder
  • Miłosz Jaguszewski
  • Michel Noutsias
  • Maike Knorr
  • Thomas Jansen
  • Wolfgang Dichtl
  • Dirk von Lewinski
  • Christof Burgdorf
  • Behrouz Kherad
  • Carsten Tschöpe
  • Annahita Sarcon
  • Jerold Shinbane
  • Lawrence Rajan
  • Guido Michels
  • Roman Pfister
  • Alessandro Cuneo
  • Claudius Jacobshagen
  • Mahir Karakas
  • Wolfgang Koenig
  • Alexander Pott
  • Philippe Meyer
  • Marco Roffi
  • Adrian Banning
  • Mathias Wolfrum
  • Florim Cuculi
  • Richard Kobza
  • Thomas A Fischer
  • Tuija Vasankari
  • K E Juhani Airaksinen
  • L Christian Napp
  • Rafal Dworakowski
  • Philip MacCarthy
  • Christoph Kaiser
  • Stefan Osswald
  • Leonarda Galiuto
  • Christina Chan
  • Paul Bridgman
  • Daniel Beug
  • Clément Delmas
  • Olivier Lairez
  • Ekaterina Gilyarova
  • Alexandra Shilova
  • Mikhail Gilyarov
  • Ibrahim El-Battrawy
  • Ibrahim Akin
  • Karolina Poledniková
  • Petr Toušek
  • David E Winchester
  • Michael Massoomi
  • Jan Galuszka
  • Christian Ukena
  • Gregor Poglajen
  • Pedro Carrilho-Ferreira
  • Christian Hauck
  • Carla Paolini
  • Claudio Bilato
  • Yoshio Kobayashi
  • Ken Kato
  • Iwao Ishibashi
  • Toshiharu Himi
  • Jehangir Din
  • Ali Al-Shammari
  • Abhiram Prasad
  • Charanjit S Rihal
  • Kan Liu
  • P Christian Schulze
  • Matteo Bianco
  • Lucas Jörg
  • Hans Rickli
  • Gonçalo Pestana
  • Thanh H Nguyen
  • Michael Böhm
  • Lars S Maier
  • Fausto J Pinto
  • Petr Widimský
  • Stephan B Felix
  • Ruediger C Braun-Dullaeus
  • Wolfgang Rottbauer
  • Gerd Hasenfuß
  • Burkert M Pieske
  • Heribert Schunkert
  • Monika Budnik
  • Grzegorz Opolski
  • Holger Thiele
  • Johann Bauersachs
  • John D Horowitz
  • Carlo Di Mario
  • Francesco Bruno
  • William Kong
  • Mayank Dalakoti
  • Yoichi Imori
  • Thomas Münzel
  • Filippo Crea
  • Thomas F Lüscher
  • Jeroen J Bax
  • Frank Ruschitzka
  • Gaetano Maria de Ferrari
  • Piero Fariselli
  • Jelena R Ghadri
  • Rodolfo Citro
  • Fabrizio D'Ascenzo
  • Christian Templin
  • InterTAK-ML

Related Research units

Abstract

AIMS: Takotsubo syndrome (TTS) is associated with a substantial rate of adverse events. We sought to design a machine learning (ML)-based model to predict the risk of in-hospital death and to perform a clustering of TTS patients to identify different risk profiles.

METHODS AND RESULTS: A ridge logistic regression-based ML model for predicting in-hospital death was developed on 3482 TTS patients from the International Takotsubo (InterTAK) Registry, randomly split in a train and an internal validation cohort (75% and 25% of the sample size, respectively) and evaluated in an external validation cohort (1037 patients). Thirty-one clinically relevant variables were included in the prediction model. Model performance represented the primary endpoint and was assessed according to area under the curve (AUC), sensitivity and specificity. As secondary endpoint, a K-medoids clustering algorithm was designed to stratify patients into phenotypic groups based on the 10 most relevant features emerging from the main model. The overall incidence of in-hospital death was 5.2%. The InterTAK-ML model showed an AUC of 0.89 (0.85-0.92), a sensitivity of 0.85 (0.78-0.95) and a specificity of 0.76 (0.74-0.79) in the internal validation cohort and an AUC of 0.82 (0.73-0.91), a sensitivity of 0.74 (0.61-0.87) and a specificity of 0.79 (0.77-0.81) in the external cohort for in-hospital death prediction. By exploiting the 10 variables showing the highest feature importance, TTS patients were clustered into six groups associated with different risks of in-hospital death (28.8% vs. 15.5% vs. 5.4% vs. 1.0.8% vs. 0.5%) which were consistent also in the external cohort.

CONCLUSION: A ML-based approach for the identification of TTS patients at risk of adverse short-term prognosis is feasible and effective. The InterTAK-ML model showed unprecedented discriminative capability for the prediction of in-hospital death.

Bibliographical data

Original languageEnglish
ISSN1388-9842
DOIs
Publication statusPublished - 12.2023

Comment Deanary

This article is protected by copyright. All rights reserved.

PubMed 37522520