Personalized diagnosis in suspected myocardial infarction
View Full Paper
Ask AI
Bookmark
Share
Why the study?
Guideline-recommended hs-cTn approaches for suspected MI require fixed assay-specific thresholds and timepoints without directly integrating clinical information.
Does the ARTEMIS machine-learning model improve diagnostic accuracy and efficiency for suspected MI compared to guideline-recommended hs-cTn algorithms in ED patients?
Population
2,575 emergency department patients with suspected MI (derivation), 1688 (validation), and 23,411 across 13 international cohorts
Comparison
ARTEMIS machine-learning models vs hs-cTn only and guideline-recommended strategy
Design
Derivation, external validation, and international generalizability study of diagnostic prediction models
Authors
Discussion
Loading...
Member takes
Implication
May enhance ED efficiency for MI rule-out; leaves open prospective validation before practice change.
Structured PICO
Does the ARTEMIS machine-learning model improve diagnostic accuracy and efficiency for suspected MI compared to guideline-recommended hs-cTn algorithms in ED patients?
P
Population
27,674 adult patients presenting to the emergency department with symptoms suggestive of myocardial infarction (excluding STEMI), median age 61, 55.8% male, multinational (9 countries).
I
Intervention
ARTEMIS diagnostic model (machine-learning based super learner) using single or serial hs-cTn measurements and 11 clinical variables to estimate individual MI probability.
C
Comparator
Guideline-recommended strategy (ESC 0/1h and 0/2h algorithms) using fixed assay-specific hs-cTn thresholds.
O
Outcome
Discriminative performance of the models assessed using area under the receiver operating characteristic curve (AUC) and logLoss.surrogate
A machine-learning model integrating clinical variables and hs-cTn concentrations safely ruled out myocardial infarction in up to three times as many patients as standard guideline algorithms.
Cite This Study
Neumann et al. (2023) studied this question.
View Full Paper
Ask AI
Bookmark
Share