Assessing Coronary Artery Disease Risk Using Seismocardiography in Patients with Chest Pain.

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Coronary artery disease (CAD) diagnosis often relies on subjective chest pain assessment, leading to variability.
• Traditional risk models may not fully leverage objective physiological data for early CAD detection.

Purpose of the Study:

• To introduce the EMR Score, a new tool integrating seismocardiography (SCG) features and clinical risk factors for estimating pre-test probability of CAD.
• To evaluate the EMR Score's performance against existing models, focusing on chest pain presence as the primary symptom.

Main Methods:

• A multicenter trial enrolled 1,640 participants (740 with obstructive CAD, 900 controls).
• Coronary artery disease (CAD) was confirmed via CCTA or ICA.
• Seismocardiography (SCG) and ECG signals were recorded; a 1D CNN model was trained using SCG features and clinical data.

Main Results:

• The EMR Score demonstrated superior performance compared to the AHA model, with a higher AUC (0.85 vs. 0.74).
• The EMR Score achieved improved specificity (42% vs. 35%) while maintaining high sensitivity (96% vs. 90%).
• It effectively reclassified intermediate-risk patients, shifting 69% to more accurate low- or high-risk categories.

Conclusions:

• The EMR Score offers a scalable, cost-effective screening tool for CAD risk stratification by integrating SCG data and clinical factors.
• By standardizing chest pain assessment and reducing interobserver variability, it enhances clinical decision-making for CAD.
• This novel approach optimizes CAD screening, particularly for diverse populations, by leveraging objective physiological signals.