Related Experiment Video
Updated: Aug 22, 2025
Identifying Coronary Artery Calcification on Non-gated Computed Tomography Scans
Published on: August 28, 2018
Mortality impact of low CAC density predominantly occurs in early atherosclerosis: explainable ML in the CAC
Fay Y Lin1, Benjamin P Goebel2, Benjamin C Lee2
1Department of Radiology, New York-Presbyterian Hospital and Weill Cornell Medicine, New York, NY, USA; Department of Population Health, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
Insights
Explainable machine learning reveals coronary artery calcium (CAC) density, particularly when very low, significantly impacts mortality risk. CAC density and characteristics improve cardiovascular mortality prediction more than diseased vessel count.
Area of Science:
- Cardiology
- Machine Learning
- Predictive Analytics
Background:
- Machine learning (ML) models for coronary artery calcium (CAC) risk prediction are effective but lack interpretability.
- Understanding the specific impact of CAC characteristics on mortality is crucial for clinical application.
Purpose of the Study:
- To use explainable ML to determine the impact and magnitude of CAC characteristics on 10-year all-cause mortality (ACM).
- To enhance the interpretability of ML models in cardiovascular risk prediction.
Main Methods:
- Trained XGBoost ML models on asymptomatic subjects using clinical data + CAC (ML 1) and additional CAC characteristics like density and number of diseased vessels (ML 2).
- Applied SHAP (SHapley Additive exPlanations) for explainable AI to analyze relationships between CAC, its characteristics, and mortality.
- Validated models on 20% of the data from the CAC consortium.
Main Results:
- ML 2 showed similar performance to ML 1 for predicting all-cause mortality (AUC 0.819 vs 0.821).
- ML 2 was superior for predicting cardiovascular (CV) mortality (AUC 0.847 vs 0.845).
- Low CAC density (≤0.75) significantly increased mortality risk, especially in patients with CAC 1-100. Number of diseased vessels did not independently increase mortality risk when CAC and density were considered.
Conclusions:
- CAC density is a critical mortality risk factor, particularly at very low levels (≤0.75), often seen in lower CAC scores.
- CAC and its density are more predictive of mortality than the number of diseased vessels.
- Explainable ML (SHAP) effectively elucidates complex relationships within predictive models, improving understanding of cardiovascular risk factors.
Background:
Machine learning (ML) models of risk prediction with coronary artery calcium (CAC) and CAC characteristics exhibit high performance, but are not inherently interpretable.
Objectives:
To determine the direction and magnitude of impact of CAC characteristics on 10-year all-cause mortality (ACM) with explainable ML.
Methods:
We analyzed asymptomatic subjects in the CAC consortium. We trained ML models on 80% and tested on 20% of the data with XGBoost, using clinical characteristics + CAC (ML 1) and additional CAC characteristics of CAC density and number of calcified vessels (ML 2). We applied SHAP, an explainable ML tool, to explore the relationship of CAC and CAC characteristics with 10-year all-cause and CV mortality.
Results:
2376 deaths occurred among 63,215 patients [68% male, median age 54 (IQR 47-61), CAC 3 (IQR 0-94.3)]. ML2 was similar to ML1 to predict all-cause mortality (Area Under the Curve (AUC) 0.819 vs 0.821, p = 0.23), but superior for CV mortality (0.847 vs 0.845, p = 0.03). Low CAC density increased mortality impact, particularly ≤0.75. Very low CAC density ≤0.75 was present in only 4.3% of the patients with measurable density, and 75% occurred in CAC1-100. The number of diseased vessels did not increase mortality overall when simultaneously accounting for CAC and CAC density.
Conclusion:
CAC density contributes to mortality risk primarily when it is very low ≤0.75, which is primarily observed in CAC 1-100. CAC and CAC density are more important for mortality prediction than the number of diseased vessels, and improve prediction of CV but not all-cause mortality. Explainable ML techniques are useful to describe granular relationships in otherwise opaque prediction models.
More Related Videos
08:51Author Spotlight: Integrated Multi-Omics Analysis for Unveiling Multicellular Immune Signatures in Clinical Heart Attack Cohorts
Published on: September 20, 2024
06:57Author Spotlight: Advancing Cardiovascular Imaging - Introducing the Spatially Weighted Calcium Score for Early Disease Detection
Published on: September 22, 2023
Related Concept Videos
01:25Imaging Studies for Cardiovascular System VI: Calcium -Scoring CT
01:30Coronary Artery Disease I: Introduction
01:26Coronary Artery Disease IV: Preventive Measures
01:26Coronary Artery Disease II: Pathophysiology
01:26Atherosclerosis III: Management
01:30Atherosclerosis I: Introduction