Machine learning for ECG diagnosis and risk stratification of occlusion myocardial infarction

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Occlusion myocardial infarction (OMI) without ST-elevation is an increasing clinical challenge.
• Patients with non-ST-elevation OMI have a poor prognosis and require timely reperfusion therapy.
• Current diagnostic tools lack accuracy for identifying these patients during initial triage.

Purpose of the Study:

• To develop and validate machine learning models for the electrocardiogram (ECG) diagnosis of OMI.
• To create an intelligent model that outperforms current diagnostic methods and clinician judgment.
• To enhance the accuracy of OMI risk assessment for improved patient management.

Main Methods:

• An observational cohort study involving 7,313 consecutive patients from multiple clinical sites.
• Development and external validation of machine learning models for ECG-based OMI diagnosis.
• Analysis of ECG features driving model performance, validated by clinical experts.

Main Results:

• The derived machine learning model demonstrated superior precision and sensitivity compared to practicing clinicians and commercial systems.
• The OMI risk score significantly improved rule-in and rule-out accuracy in routine care settings.
• Integration with clinical judgment reclassified one in three chest pain patients correctly.

Conclusions:

• Machine learning models can effectively diagnose OMI in patients with no ST-elevation on ECG.
• This intelligent tool offers a significant advancement in early OMI detection and management.
• Validated ECG features provide mechanistic insights into myocardial injury, supporting clinical application.