EUVIMEDEUVIMED entdecken
Systematische ÜbersichtMetaanalyseOpen Access · CC BY 4.0

Accuracy of Machine Learning Models for Early Prediction of Major Cardiovascular Events Post Myocardial Infarction: A Systematic Review and Meta-Analysis

Genauigkeit von Machine-Learning-Modellen zur frühen Vorhersage schwerer kardiovaskulärer Ereignisse nach Myokardinfarkt: systematische Übersicht und Metaanalyse

Yi Xiang iD, Dong Liu iD, Leilei Guo iD, Yuhua Zheng iD, Xiaoman Xiong iD, Tao Xu iD

Reviews in Cardiovascular Medicine · 17.06.2025 · 26 (6)

DOI: 10.31083/rcm37224 ↗

Auf einen Blick

28 Studien untersuchten ML-Modelle zur Vorhersage von MACEs nach Myokardinfarkt. In den Validierungssets lag der gepoolte C-Index bei 0,77.

Fragestellung

Wie genau sagen Machine-Learning-Modelle schwere kardiovaskuläre Ereignisse nach Myokardinfarkt vorher?

Methode

Systematische Literaturrecherche in Cochrane, Embase, PubMed und Web of Science bis 19. Juni 2024; PROBAST-Bewertung; Subgruppenanalyse nach PCI.

Ergebnisse

28 Studien mit 59.392 Patient:innen mit Myokardinfarkt. Validierungssets: C-Index 0,77 (95 %-KI 0,74–0,81), Sensitivität 0,78 (95 %-KI 0,73–0,82), Spezifität 0,85 (95 %-KI 0,81–0,89). Bei PCI: C-Index 0,73 (95 %-KI 0,66–0,79).

Schlussfolgerung der Autor:innen

ML-Modelle auf Basis klinischer Merkmale nach Myokardinfarkt beeinflussen die Vorhersagegenauigkeit. Zukünftige Studien sollten größere Stichproben und vereinfachte Werkzeuge prüfen.

Kernaussagen

  • 28 Studien mit 59.392 Patient:innen wurden eingeschlossen
  • Validierungssets: C-Index 0,77; Sensitivität 0,78; Spezifität 0,85
  • Bei PCI lag der C-Index bei 0,73

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Systematische Übersicht in der Zeitschrift Reviews in Cardiovascular Medicine, online erschienen am 17.06.2025. Verfasst von Yi Xiang und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Postgraduate Students, Guizhou University of Traditional Chinese Medicine, 550000 Guiyang, Guizhou, China. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Key Laboratory of Translational Medicine for the Prevention and Treatment of Diseases with Integrated Traditional Chinese and Western Medicine in Guizhou Provi…, National Natural Science Foundation of China.

Auf einen BlickAbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Major adverse cardiovascular events (MACEs) significantly affect the prognosis of patients with myocardial infarction (MI). With the widespread application of machine learning (ML), researchers have attempted to develop models for predicting MACEs following MI. However, there remains a lack of evidence-based proof to validate their value. Thus, we conducted this study to review the ML models’ performance in predicting MACEs following MI, contributing to the evidence base for the application of clinical prediction tools.

Methods

A systematic literature search spanned four major databases (Cochrane, Embase, PubMed, Web of Science) with entries through to June 19, 2024. With the Prediction Model Risk of Bias Assessment Tool (PROBAST), the risk of bias in the included models was appraised. Subgroup analyses based on whether patients had percutaneous coronary intervention (PCI) were carried out for the analysis.

Results

Twenty-eight studies were included for analysis, covering 59,392 patients with MI. The pooled C-index for ML models in the validation sets was 0.77 (95% CI 0.74–0.81) in predicting MACEs post MI, with a sensitivity (SEN) and specificity (SPE) of 0.78 (95% CI 0.73–0.82) and 0.85 (95% CI 0.81–0.89), respectively; the pooled C-index was 0.73 (95% CI 0.66–0.79) in the validation sets, with an SEN of 0.75 (95% CI 0.67–0.81) and an SPE of 0.84 (95% CI 0.75–0.90) in patients who underwent PCI. Logistic regression was the predominant model in the studies and demonstrated relatively high accuracy.

Conclusions

ML models based on clinical characteristics following MI, influence the accuracy of prediction. Therefore, future studies can include larger sample sizes and develop simplified tools for predicting MACEs. The PROSPERO registration: CRD42024564550, https://www.crd.york.ac.uk/PROSPERO/view/CRD42024564550.

Schlagwörter

Stichwörter der Autor:innen

MACEsPCImachine learningmyocardial infarction

Autor:innen und Institutionen

  • School of Postgraduate Students, Guizhou University of Traditional Chinese Medicine, 550000 Guiyang, Guizhou, China.
  • School of Medicine, Guizhou University of Traditional Chinese Medicine, 550000 Guiyang, Guizhou, China.
  • Cardiovascular Medicine, The Second Affiliated Hospital of Guizhou University of Traditional Chinese Medicine, 550000 Guiyang, Guizhou, China.

ORCID-Profile: Yi Xiang, Dong Liu, Leilei Guo, Yuhua Zheng, Xiaoman Xiong, Tao Xu

Förderung

  • Key Laboratory of Translational Medicine for the Prevention and Treatment of Diseases with Integrated Traditional Chinese and Western Medicine in Guizhou Provi…
  • National Natural Science Foundation of China

Zitieren

APA

Xiang, Y., Liu, D., Guo, L., Zheng, Y., Xiong, X., & Xu, T. (2025). Accuracy of Machine Learning Models for Early Prediction of Major Cardiovascular Events Post Myocardial Infarction: A Systematic Review and Meta-Analysis. Reviews in Cardiovascular Medicine, 26(6). https://doi.org/10.31083/rcm37224

Vancouver

Xiang Y, Liu D, Guo L, Zheng Y, Xiong X, Xu T. Accuracy of Machine Learning Models for Early Prediction of Major Cardiovascular Events Post Myocardial Infarction: A Systematic Review and Meta-Analysis. Reviews in Cardiovascular Medicine. 2025;26(6). doi:10.31083/rcm37224

BibTeX
@article{Xiang2025,
  author = {Xiang, Yi and Liu, Dong and Guo, Leilei and Zheng, Yuhua and Xiong, Xiaoman and Xu, Tao},
  title = {Accuracy of Machine Learning Models for Early Prediction of Major Cardiovascular Events Post Myocardial Infarction: A Systematic Review and Meta-Analysis},
  journal = {Reviews in Cardiovascular Medicine},
  year = {2025},
  volume = {26},
  number = {6},
  doi = {10.31083/rcm37224}
}
Kostenlos starten →