Artificial Intelligence-Enhanced Electrocardiography and Health Records to Predict Cardiac Arrest

Area of Science:

• Cardiology
• Artificial Intelligence
• Public Health

Background:

• Out-of-hospital cardiac arrest (OHCA) presents a significant public health challenge.
• Current strategies for predicting OHCA risk in the general population are limited.

Purpose of the Study:

• To evaluate the effectiveness of artificial intelligence (AI)-enhanced electrocardiography (ECG) and electronic health records (EHRs) in stratifying OHCA risk.
• To assess the real-world utility of AI models for predicting OHCA incidence.

Main Methods:

• A case-control study design was employed for model derivation and temporal validation.
• The study evaluated the 2-year cumulative incidence of OHCA in individuals undergoing ECG, accounting for competing mortality risks.

Main Results:

• The multimodal AI-enhanced ECG + EHR model demonstrated the highest discrimination for OHCA (AUC: 0.83).
• In a real-world cohort, the AI model identified two-thirds of individuals who experienced OHCA over two years.
• High-risk individuals identified by the model had a 2.4% 2-year cumulative incidence of OHCA, compared to 0.5% in low-risk individuals.

Conclusions:

• AI-enhanced ECG and EHR data effectively identify individuals at risk of OHCA.
• The models provide clinically relevant risk stratification for incident OHCA over a 2-year period within a large healthcare system.