Predicting Occlusion Myocardial Infarctions in the Emergency Department Using Artificial Intelligence

Area of Science:

• Cardiology
• Artificial Intelligence in Medicine
• Emergency Medicine

Background:

• Acute coronary occlusion myocardial infarction (OMI) is a critical condition requiring rapid diagnosis and intervention.
• Current diagnostic criteria, such as ST-elevation myocardial infarction (STEMI), may miss a significant proportion of OMI cases.
• Timely intervention for OMI is often delayed, contributing to adverse patient outcomes.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for predicting OMI in emergency department (ED) patients presenting with chest pain.
• To utilize readily available early ED assessment data, including electrocardiogram (ECG), medical history, and initial lab values.
• To compare the AI model's performance against established STEMI criteria.

Main Methods:

• A deep-learning AI model was developed using data from 24,511 adult ED patients with chest pain across 5 Swedish hospitals.
• OMI cases were identified through register data and health record review; STEMI cases bypassing the ED were excluded.
• The AI model was trained on ECG data, optionally combined with other early ED information, and internally validated.

Main Results:

• The AI model achieved a high area under the receiver operating characteristic (AUC) of 95.3% for OMI prediction.
• At 97.4% specificity, the AI model demonstrated a sensitivity of 62%, significantly higher than the 27% sensitivity of STEMI criteria.
• Only 5.4% of identified OMI cases received timely angiography, highlighting a critical gap in current care pathways.

Conclusions:

• The developed AI model shows excellent performance in identifying OMI in ED chest pain patients, outperforming STEMI criteria in sensitivity.
• Implementation of this AI model could expedite diagnosis and reduce delays in initiating interventions for OMI.
• The findings suggest a potential to improve patient outcomes by addressing the current delays in OMI treatment.