PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 10, 2026Resuscitation1 citationsOpen Access

Phenotypic clustering of myocardial infarction complicated by out-of-hospital cardiac arrest using unsupervised machine learning

MSManveer SinghGovernment Medical CollegeACAlain CariouInsermOVOlivier VarenneDélégation Paris 5

Key Result

Unsupervised machine learning identified three phenotypes among MI-related OHCA patients with 90-day mortality of 22.5%, 53.0%, and 77.2% for clusters 1, 2, and 3 respectively; cluster 2 had HR 2.97 and cluster 3 HR 6.75 for mortality versus cluster 1 (p<0.001).

Key Points

  • The research aims to classify patients with myocardium infarction (MI) and out-of-hospital cardiac arrest (OHCA) into distinct phenotypes for better management.
  • Utilized unsupervised machine learning techniques
  • Analyzed patient data related to myocardial infarction and OHCA
  • Identified different phenotypes based on outcomes
  • Discovered three distinct phenotypes among patients
  • Each phenotype associated with unique outcomes
  • Classification may improve personalized management and prognosis

Study Design

Type

Observational (n=478)

Multicenter

No

Structured PICO

P
Population
478 adult patients (≥18 years) admitted to the intensive care unit after out-of-hospital cardiac arrest (OHCA) complicating acute myocardial infarction, who achieved return of spontaneous circulation and underwent emergency invasive coronary angiography and percutaneous coronary intervention (PCI). Mean age 60.4 years, 78.9% male, single-center (France).
O
Outcome
In-hospital outcomes including all-cause mortality, bleeding complications (BARC 3-5), definite or probable stent thrombosis, and neurological outcome (CPC scale), as well as 90-day all-cause mortality.hard clinical

Unsupervised machine learning identified three distinct clinical phenotypes in patients with MI-related OHCA that strongly correlate with in-hospital and 90-day mortality, enabling better risk stratification.

Main Result

Effect estimate: HR 2.97 for cluster 2 vs cluster 1; HR 6.75 for cluster 3 vs cluster 1 (95% CI 95% CI 2.07-4.26 for cluster 2; 4.74-9.60 for cluster 3)

p-value: p=<0.001

Limitations

  • Single-center study limiting generalizability
  • Observational design with potential confounding between cluster assignment and management
  • Cause of death not captured, limiting insight into death etiology
  • Clustering based on admission variables only, dynamic changes during hospitalization not included
  • External validation in independent cohorts needed

Abstract

Unsupervised machine learning identified three phenotypes among patients with MI-related OHCA associated with distinct outcomes. This phenotypic classification may facilitate personalized management and refined prognostic assessment in this high-risk population.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Singh et al. (2026) conducted an observational in Adults with myocardial infarction complicated by out-of-hospital cardiac arrest admitted to ICU after return of spontaneous circulation and emergency invasive coronary angiography (n=478). Unsupervised machine learning clustering of admission clinical and laboratory data vs. No clustering (comparison between clusters) was evaluated on 90-day all-cause mortality (HR 2.97 for cluster 2 vs cluster 1; HR 6.75 for cluster 3 vs cluster 1, 95% CI 95% CI 2.07-4.26 for cluster 2; 4.74-9.60 for cluster 3, p=<0.001). Unsupervised machine learning identified three phenotypes among MI-related OHCA patients with 90-day mortality of 22.5%, 53.0%, and 77.2% for clusters 1, 2, and 3 respectively; cluster 2 had HR 2.97 and cluster 3 HR 6.75 for mortality versus cluster 1 (p<0.001).

synapsesocial.com/papers/69af944f70916d39fea4b516https://doi.org/10.1016/j.resuscitation.2026.111037
Ask AI
Helpful
Bookmark
Share
View Full Paper