AI-Estimated Accelerated Biological Aging on Chest X-ray Predicts Mortality: Study

A large Korean study published in the journal of Radiology Artificial Intelligence found that accelerated biological aging, as estimated from chest X-rays using a deep-learning model, is significantly associated with increased patient mortality, which suggests its potential role as a prognostic marker.

This research analyzed chest X-rays from more than 421,000 Korean adults collected between 2006 and 2020 to determine whether “radiographic age” (AI-derived estimate) could predict mortality. Using a deep learning model known as AgeNet, which was trained on healthy individuals, this study compared the radiographic age of each person with their actual chronological age.

The results suggest that when the body appears older than its actual age on imaging, it may signal significantly higher health risks. The participants whose radiographic age exceeded their chronological age by 5 or more years were classified as experiencing “accelerated aging.” Across the median follow-up period of 8.5 years, over 6,500 deaths were recorded, which included those from cardiovascular disease, cancer, and respiratory illness.

Accelerated aging was strongly associated with increased mortality across all causes and this effect was more pronounced in women. Men with accelerated aging had a 26% increased risk of death, while women faced a 52% higher risk, which highlighted the potential sex-based differences in how aging impacts health.

This study also examined how quickly radiographic age (aging velocity) changes over time. Among nearly 180,000 individuals who had at least 3 chest X-rays, those with faster increases in radiographic age underwent significantly higher mortality risks, regardless of their initial health status.

Each standard deviation increase in aging velocity corresponded to a 24% rise in mortality risk for men and 35% for women. Slower aging rates were linked to better outcomes, where women with decelerated aging velocity (<0.5 years of radiographic aging per year) experienced around 50% reduction in mortality risk.

The individuals whose radiographic age increased rapidly (>1.5 years) were at markedly higher risk of death. Mortality rates rose by 51% in men and 71% in women within this group. Overall, the study found that both accelerated radiographic aging and rapid aging velocity independently predict mortality risk. These findings suggest that AI-enhanced analysis of routine chest X-rays could become a valuable tool in preventive medicine.

Reference:

Chang, Y., Kim, H., Lee, S., Lee, H., Yoon, S. H., & Ryu, S. (2026). Accelerated aging and aging velocity from deep learning-based chest radiograph-derived age for predicting cause-specific mortality. Radiology. Artificial Intelligence, e250609, e250609. https://doi.org/10.1148/ryai.250609