Machine learning-derived aging index predicts risk in cardiovascular-kidney-metabolic syndrome
Published in Artificial Intelligence, Diseases!, Journals, Personalized medicine.
Publication date: October 10, 2026.
A retrospective study published in PLOS Medicine on October 6 introduces CKMAI, an aging index designed for cardiovascular-kidney-metabolic syndrome. Researchers used data from 6,896 adults in the US NHANES surveys from 2005–2018 to develop and evaluate the index.
The machine learning framework compared more than 100 candidate survival models and selected a simple Cox model. The resulting index uses 18 routine blood-test and clinical measures. Its reported area under the curve for five-year all-cause mortality was 0.907, outperforming comparison indices in the study.
Researchers also assessed the model across time periods within NHANES and conducted preliminary testing in a Chinese hospital cohort of 261 patients for identifying high-risk CKM status. That smaller cohort lacked complete long-term mortality follow-up.
The findings concern statistical prediction, rather than an intervention that slows aging or prevents death. Independent validation of mortality prediction in larger, geographically diverse cohorts remains necessary before generalizing its performance. A high discrimination score does not establish that using the index improves patient outcomes.