Machine Learning to Predict the Likelihood of Acute Myocardial Infarction
Why the study?
Diagnostic approaches for suspected myocardial infarction do not account for variations in cardiac troponin concentrations by age, sex, and sampling time.
Does a machine learning algorithm (MI3) improve the prediction of acute myocardial infarction in patients with suspected myocardial infarction compared to standard ESC pathways?
Comparison
MI3 algorithm vs 99th percentile and European Society of Cardiology rule-out pathways
Design
Machine learning model development and validation study
Authors
Discussion
Loading...
Member takes
Implication
MI3 may improve AMI risk stratification over ESC pathways in cohorts; leaves open clinical impact pending prospective validation.
Structured PICO
Does a machine learning algorithm (MI3) improve the prediction of acute myocardial infarction in patients with suspected myocardial infarction compared to standard ESC pathways?
P
Population
11,011 patients with suspected myocardial infarction (3,013 in training set, 7,998 in test set)
I
Intervention
Machine learning algorithm (myocardial-ischemic-injury-index [MI3]) incorporating age, sex, and paired high-sensitivity cardiac troponin I concentrations
C
Comparator
European Society of Cardiology 0/3-hour pathway and the 99th percentile at any time point
O
Outcome
Diagnosis of type 1 myocardial infarction assessed by calibration and area under the receiver operating characteristic curve (AUC)surrogate
A machine learning algorithm incorporating age, sex, and paired high-sensitivity troponin I concentrations provides highly accurate, individualized risk assessment for acute myocardial infarction, outperforming standard ESC rule-out pathways.
Cite This Study
Than et al. (2019) studied this question.