AI-Driven Aging Clocks Reveal Links Between Metabolism and Lifespan

Researchers at King’s College London have made significant advancements in the field of biological aging by developing AI-driven metabolomic aging clocks, which analyze blood metabolite data to predict health outcomes and lifespan.

This research provides insights into the relationship between biological age and health risks, offering valuable guidance for preventive health measures.

Study Overview

The study, conducted by the Institute of Psychiatry, Psychology & Neuroscience at King’s College London, evaluated the effectiveness of these AI-based aging clocks.

The team utilized data from over 225,000 participants in the UK Biobank, aged between 40 and 69, and trained and assessed 17 different machine learning algorithms to determine their accuracy in estimating lifespan and correlating with various health and aging indicators.

One crucial concept emerging from the study is “MileAge,” which refers to a person’s biological age derived from their blood metabolites—small molecules generated during metabolic processes.

The difference between a person’s predicted biologically derived age (MileAge) and their chronological age is known as the MileAge delta.

This delta serves as an indicator of whether a person’s biological aging is progressing more quickly or more slowly.

Key Findings

The findings from this research, published in Science Advances, represent a landmark comparison of various machine learning algorithms aimed at creating biological aging clocks using metabolite data, based on one of the largest health datasets available worldwide.

The study received support from the National Institute for Health and Care Research Maudsley Biomedical Research Centre.

Results indicated that people showing signs of accelerated aging—those with a MileAge greater than their chronological age—were more likely to experience increased frailty, a greater risk of chronic diseases, poorer self-reported health, and a higher risk of mortality.

Furthermore, these people had shorter telomeres, which are biological markers of cellular aging linked to age-related conditions like atherosclerosis.

Conversely, a slower rate of biological aging, indicated by a MileAge younger than chronological age, was only weakly connected to better health outcomes.

These aging clocks hold great promise for identifying early signs of declining health, allowing for more targeted preventive care before diseases manifest.

They also enable people to take a proactive approach to their health by encouraging informed lifestyle choices that could potentially enhance longevity.

Implications for Future Research

Expert commentary on the research highlights the transformative potential of metabolomic aging clocks.

The lead author emphasized that these tools could be pivotal in identifying people at higher risk for health complications as they age.

Unlike chronological age, which cannot be changed, biological age can be influenced, providing a valuable metric for health research that can inform personal lifestyle decisions and public health strategies.

The study also noted that non-linear algorithms, particularly Cubist regression, performed best in detecting aging signals.

The senior author stressed the importance of advanced analytics in big data to improve the accuracy of these tools, noting that this research marks a significant step forward in understanding the role of biological aging clocks in health decision-making.

The research identified Cubist rule-based regression as the most effective machine learning algorithm for linking biological aging with health indicators.

Additionally, algorithms capable of modeling non-linear relationships between metabolites and age demonstrated superior performance in capturing indicators related to health and lifespan.

For further reading, more details can be found in the study titled “Metabolomic age (MileAge) predicts health and life span: A comparison of multiple machine learning algorithms,” published on December 18, 2024, in Science Advances.

This initiative was funded by the National Institute for Health Research.

Study Details:

• Title: Metabolomic age (MileAge) predicts health and life span: A comparison of multiple machine learning algorithms
• Authors: Julian Mutz, Raquel Iniesta, and Cathryn M. Lewis
• Journal: Science Advances
• Publication Date: 18 December 2024
• DOI: 10.1126/sciadv.adp3743