AI Detects Early Signs of Aging from Chest X-Rays

Chronological age does not always reflect how fast the body is truly aging, and current biological age tests often rely on DNA-based markers that may miss early organ-level decline. Detecting subtle, age-related changes before symptoms appear is critical for preventing cardiovascular and lung disease. A new study now shows that artificial intelligence (AI) can extract aging signals directly from standard chest X-rays, offering a more sensitive way to capture early cardiopulmonary aging than widely used epigenetic clocks.

For the study, investigators at the Hinda and Arthur Marcus Institute for Aging Research, Boston, Massachusetts, USA), along with collaborators, used data from the U.S.-based Project Baseline Health Study. The team developed a deep learning model that analyzes routine chest X-rays to estimate biological age based on subtle structural and tissue-level changes. Rather than relying on DNA methylation patterns, the model evaluates age-related features in the heart, lungs, and surrounding anatomy visible on imaging. This allows it to capture functional and structural aging processes that may not yet be reflected in molecular biomarkers.

The AI-derived metric, known as CXR-Age, was compared with two established DNA-based epigenetic clocks, Horvath Age and DNAm PhenoAge. Researchers assessed how well each method correlated with early markers of aging-related disease, focusing on cardiopulmonary health and systemic indicators of decline. By using a widely available and low-cost imaging test, the approach has the potential to integrate seamlessly into routine clinical care without additional invasive procedures or specialized laboratory testing.

Researchers analyzed data from 2,097 adults enrolled in the Project Baseline Health Study and evaluated associations between biological age measures and subclinical disease markers. CXR-Age showed strong links to coronary artery calcium, reduced lung function, increased frailty, and higher levels of proteins associated with inflammation and aging. In contrast, the epigenetic clocks demonstrated weaker or inconsistent associations, particularly in middle-aged participants. This suggests that imaging-based AI may capture earlier and more clinically relevant signs of organ aging.

The findings, published in The Journals of Gerontology, indicate that AI analysis of chest X-rays may serve as a more practical and sensitive indicator of cardiopulmonary aging than existing DNA-based methods. Such tools could help identify individuals at higher risk of age-related disease before clinical symptoms emerge. Researchers suggest that this approach could complement traditional risk assessments and support more personalized, preventive healthcare strategies. Future work will explore how CXR-Age performs across diverse populations and whether it can guide early intervention decisions.

“These findings suggest that deep learning applied to common medical images can reveal how our organs are aging — information that might one day help clinicians identify people at risk of age-related disease before symptoms develop,” said Douglas P. Kiel, MD, MPH, co-author of the study. “AI tools like this could become an important complement to traditional risk assessments.”

Related Links:
Hinda and Arthur Marcus Institute for Aging Research