Annals of Internal MedicineDeep Learning to Estimate Cardiovascular Risk From Chest Radiographs

A Risk Prediction Study

In an era where precision and efficiency are paramount in healthcare, the integration of artificial intelligence into diagnostic processes marks a significant leap forward. The realm of deep learning, utilizing chest radiographs (CXR) to estimate the 10-year risk for major adverse cardiovascular events (MACE), offers a novel perspective on cardiovascular risk assessment. This approach not only complements but potentially surpasses traditional methods, providing a new avenue for clinicians to identify high-risk patients, particularly when conventional risk factors are elusive.

Key Points:

• A deep-learning model, CXR CVD-Risk, was developed to estimate the 10-year risk for major adverse cardiovascular events (MACE) using routine chest radiographs (CXRs), presenting a novel tool for cardiovascular risk assessment.
• The model was externally validated in two distinct outpatient groups: one with unknown ASCVD risk due to missing conventional risk factors and another with calculable ASCVD risk scores, demonstrating its applicability across varied clinical scenarios.
• In outpatients with indeterminable ASCVD risk, the CXR CVD-Risk model effectively identified individuals with a 7.5% or higher 10-year MACE risk, indicating its potential to fill gaps left by traditional risk assessment methods.
• Comparison with the established ASCVD risk score revealed that CXR CVD-Risk provided additional predictive value for MACE, suggesting its utility as a supplementary tool or a standalone predictor when conventional inputs are unavailable.
• The model’s predictive power was underscored by its ability to forecast MACE with greater accuracy than the ASCVD risk score alone in patients with known risk factors, emphasizing its potential in enhancing personalized medicine.
• Despite its promising results, the retrospective nature of the study and its reliance on electronic medical records highlight the need for prospective validation to establish its real-world efficacy and integration into clinical practice.

HCN Medical Memo
CXR CVD-Risk predicts 10-year MACE beyond the clinical standard, offering a new paradigm in identifying high-risk individuals.

More in Cardiology