# First-line risk stratification with machine learning models facilitates rapid triage for non-ST-elevation myocardial infarction

## TL;DR

A machine learning model using initial blood tests and patient data can quickly and accurately identify heart attack risk in emergency patients.

## Contribution

A novel ML-based risk stratification model for NSTEMI that improves triage speed and accuracy compared to standard troponin testing.

## Key findings

- The ML model outperformed high-sensitivity troponin testing alone in predicting NSTEMI.
- Combining the ML model with existing guidelines safely ruled in/out 85.3% of patients within 1 hour.
- The model achieved 98.8% negative predictive value for low-risk patients and 78.1% positive predictive value for high-risk patients.

## Abstract

Timely diagnosis of non-ST-elevation myocardial infarction (NSTEMI) remains challenging, as current protocols rely on serial high-sensitivity cardiac troponin (hs-cTn) tests that may delay decisions and overcrowd emergency departments. We retrospectively analyzed 54,636 patients receiving hs-cTn testing at emergency departments across Taiwan (May 2016–Dec 2021). Excluding STEMI and incomplete cases, we developed a machine learning (ML) model using demographics and 23 routine lab tests from the initial blood draw to enable early NSTEMI risk stratification. An actionable clinical decision supporting algorithm was also created based on ML-derived risk scores. A total of 15,096 eligible patients (mean age 69.94 ± 15.66 years; 42.2% female) were included in model training and evaluation. The ML model outperformed hs-cTn alone in both internal and external validation sets in terms of area under the receiver-operating characteristic curve. Beyond model development, a clinically actionable decision algorithm using risk score was established. Thresholds (<1.8 and ≥38.5) to define low- and high-risk groups, the model achieved a negative predictive value (NPV) of 98.8% (98.5–99.1%) for rule-out and a positive predictive value (PPV) of 78.1% (73.2–82.4%) for rule-in, encompassing 48.3% and 2.6% of patients, respectively. When combined with the established 0 h/1 h algorithm, the ML model further enhanced early decision-making, safely ruling in/out 85.3% of patients within 1 hour, with PPV and NPV reaching 84.9% (79.5–87.7%) and 100% (99.6–100%), respectively. In conclusion, this ML-based approach offers not only accurate prediction but also an actionable guide to support rapid, safe NSTEMI triage in emergency care.

Every minute counts when someone comes to the emergency room with chest pain, yet identifying a heart attack quickly and accurately remains a major challenge, especially when early test results are inconclusive. In our study, we wanted to improve the initial evaluation of people suspected of having a certain type of heart attack, called non-ST-elevation myocardial infarction, or NSTEMI. We developed computer models that use routine blood tests and basic patient information, collected right when the patient arrives, to predict their risk of having a heart attack. These models worked well, even better than relying on standard blood tests alone, and could identify over half of the patients as either low or high risk within the first blood draw. We also found that combining our model with existing guidelines made early triage faster and more accurate. Our approach could help emergency doctors make quicker decisions, reduce unnecessary waiting, and prioritize care for those who need it most. We believe this kind of machine learning tool, based on real-world data and simple tests, could be a practical step forward in emergency heart care.