Clinical Scorecard: Utilizing a Machine Learning-Enhanced Magnetocardiography Model to Forecast Angina Risk Following Percutaneous Coronary Intervention

At a Glance

Key Highlights

• MCG is a noninvasive, radiation-free modality sensitive to ischemia-induced ionic current changes, enabling functional cardiac assessment post-PCI.
• A combined model integrating MCG scores with serum biomarkers (LDL-C, cTnI, NT-proBNP) improves prediction accuracy for post-PCI angina.
• Machine learning-based MCG models demonstrated good discrimination (AUC >0.8) and calibration for predicting angina stability and frequency after PCI.

Guideline-Based Recommendations

Diagnosis

• Use magnetocardiography (MCG) pre- and within 72 hours post-PCI to derive quantitative scores reflecting cardiac electrical activity.
• Assess angina status within 3 months post-PCI using validated patient-reported outcome measures such as the Seattle Angina Questionnaire (SAQ) Angina Stability and Angina Frequency domains.
• Incorporate serum biomarkers including LDL-C, cardiac troponin I, and NT-proBNP to enhance risk stratification.

Management

• Apply machine learning-based MCG models alone or combined with biomarkers to identify patients at high risk for recurrent angina after PCI.
• Use risk stratification outputs (e.g., nomograms) to guide clinical decision-making and tailor follow-up strategies.

Monitoring & Follow-up

• Perform serial MCG assessments as a noninvasive, rapid, and radiation-free method suitable for repeated evaluations post-PCI.
• Monitor angina symptoms using standardized questionnaires like SAQ to track patient-reported outcomes over time.

Risks

• Recognize limitations of traditional modalities (coronary angiography, CMR, SPECT, echocardiography) including invasiveness, radiation exposure, contrast contraindications, and operator dependency.
• Consider MCG as a safer alternative for repeated functional cardiac assessments, especially in patients with renal impairment or contraindications to contrast agents.

Patient & Prescribing Data

110 patients with coronary artery disease undergoing successful PCI, excluding those with arrhythmias, severe heart failure, or poor-quality MCG data

MCG scores decreased significantly post-PCI; combined MCG and biomarker models showed improved sensitivity, F1-score, and AUC for predicting angina outcomes, supporting their use in clinical risk stratification.

Clinical Best Practices

• Integrate machine learning-enhanced MCG with clinical biomarkers for comprehensive risk assessment of post-PCI angina.
• Utilize validated patient-reported outcome tools such as the Seattle Angina Questionnaire for symptom evaluation.
• Employ noninvasive, rapid, and radiation-free MCG for serial monitoring to detect residual ischemia and guide management.
• Exclude patients with arrhythmias or poor-quality MCG recordings to ensure data reliability.
• Use nomograms derived from combined models to facilitate individualized risk prediction and clinical decision-making.

References

• Seattle Angina Questionnaire validation and use
• Magnetocardiography principles and clinical applications
• Limitations of conventional post-PCI assessment modalities