Heart scan analysis improved heart failure risk prediction
For 10-year heart failure prediction, a scan-based model scored 0.84 versus 0.72 for a clinical score in distinguishing future cases from people who remained unaffected.
Our read
Biomarkers- Studied:
- Cohort study of 4,182 people without symptoms or known cardiac disease, mean 3.9 years
- What changed:
- Area under the curve for 10-year heart failure prediction was higher with radiomics than PREVENT-HF: 0.84 versus 0.72
- The catch:
- Average follow-up was 3.9 years despite the 10-year prediction estimates
- Why it’s aging research:
- It bears on staying healthy in midlife and later life.
Editor of record: Boris Djordjevic · AgingResearch.News
Heart scans from 5,969 people without symptoms or known cardiac disease formed the basis of a retrospective observational study of heart failure prediction. These computed tomography scans used X-rays to assess calcium in the arteries supplying the heart. A computer model analyzed image patterns and shapes of heart chambers and nearby blood vessels.
The independent testing group included 4,182 participants with an average age of 58 years; 45% were women. During an average 3.9 years of follow-up, 263 heart failure events were recorded. For 10-year predictions, the scan-based model scored 0.84 versus 0.72 for PREVENT-HF, a clinical risk score, on a measure of how well predictions distinguished future cases from people who remained unaffected. Combining both models raised that score to 0.85.
Why it matters
Heart failure risk assessment based mainly on clinical information may miss structural changes before symptoms appear. The authors suggest routine scans may capture such changes associated with later heart failure.
Caveats
Heart failure outcomes came from diagnostic codes, and average follow-up was 3.9 years despite the 10-year prediction estimates. The findings concern people without symptoms or known cardiac disease and were reported in a conference abstract.
The paper
CAC CT-Based Cardiac Radiomics for Heart Failure Outcome Prediction
Emory Univ
AHA Scientific Sessions 2026 Abstract, Nov 2, 2026
Presented at AHA Scientific Sessions 2026, abstract 14618
Comments
Be the first to comment.