Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
December 9, 2025Revista de Investigación e Innovación en Ciencias de la SaludOpen Access

AI models achieve an AUC of ~0.82 for diagnosing major cardiovascular conditions.

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Cardiovascular disease poses a substantial global health burden with diagnostic delays compromising outcomes, prompting evaluation of the diagnostic effectiveness of artificial intelligence models across major cardiovascular conditions.

Do artificial intelligence models provide effective diagnostic performance for major cardiovascular conditions?

Population

Thirty-five studies evaluating artificial intelligence models for major cardiovascular conditions

Design

Systematic review and meta-analysis

Key result

Artificial intelligence models demonstrated a pooled area under the receiver operating characteristic curve of 0.823 (95% CI 0.754-0.892) for diagnosing major cardiovascular conditions.

Authors

CZCarlos Zepeda-LugoMSMarcos Alberto Sanchez-LizarragaAIAndrea Insfran-Rivarola

Discussion

Loading...

Member takes

Overview

AI models show promising discrimination for CAD, ACS/MI, and HF; leaves open routine adoption pending external validation and local calibration.

Study Design

Type

Meta-Analysis (n=35)

Structured PICO

Do artificial intelligence models provide effective diagnostic performance for major cardiovascular conditions?

P
Population
35 studies evaluating the diagnostic effectiveness of artificial intelligence models for major cardiovascular conditions.
I
Intervention
Artificial intelligence models for cardiovascular disease diagnosis
O
Outcome
Diagnostic performance measured by pooled area under the receiver operating characteristic curve (AUC)surrogate

Main Result

Effect estimate: Pooled AUC 0.823 (95% CI 0.754-0.892)

AI models show promising diagnostic discrimination for cardiovascular diseases, but high heterogeneity and lack of external validation highlight the need for local calibration before clinical implementation.

Limitations

  • Considerable heterogeneity (I² = 98.4%)
  • Only 9 of 35 included studies reported external/independent validation
  • Variable specificity and predictive value

Cite This Study

Zepeda-Lugo et al. (2025) conducted a meta-analysis in Cardiovascular disease (coronary artery disease, acute coronary syndromes, myocardial infarction, and heart failure) (n=35). Artificial intelligence models was evaluated on Diagnostic performance (pooled area under the receiver operating characteristic curve) (Pooled AUC 0.823, 95% CI 0.754-0.892). Artificial intelligence models demonstrated a pooled area under the receiver operating characteristic curve of 0.823 (95% CI 0.754-0.892) for diagnosing major cardiovascular conditions.

synapsesocial.com/papers/6aa558683f91f0dc5278eb31https://doi.org/10.46634/riics.504
View Full Paper
Ask AI
Bookmark
Share