Researchers at the Institute of Psychiatry, Psychology & Neuroscience (IoPPN) at King’s College London have made significant strides in the application of artificial intelligence (AI) to the field of healthcare, particularly in predicting human biological age and lifespan. Their recent study, published in Science Advances, meticulously evaluated a variety of machine learning algorithms to establish their effectiveness in determining biological age based on metabolic data derived from human blood samples.
Dr. Julian Mutz, the lead author of the study, along with co-authors Raquel Iniesta and Cathryn M. Lewis, emphasised the comprehensive nature of their research, stating, “This study presents a comprehensive comparison of machine learning algorithms for developing metabolomic aging clocks, benchmarking a wide range of models under consistent conditions in one of the largest metabolomics datasets available globally.” The findings of this research are particularly noteworthy as the adoption of digital health and AI tools continues to rise globally amongst healthcare professionals.
Metabolomics, the study of molecules produced through metabolism, plays a central role in this research. Metabolism itself is broadly categorised into two types: catabolism, which involves the breakdown of substances for energy, and anabolism, which deals with the construction of complex molecules. The research team focused on analysing plasma metabolite data obtained from over 225,000 individuals participating in the UK Biobank, with a mean age of approximately 57 years.
Using nuclear magnetic resonance (NMR) spectroscopy, a noninvasive technique for chemical analysis, the researchers examined 168 metabolites from the blood of these participants. They aimed to assess the performance of 17 different AI algorithms in predicting life span based on metabolic data. Through their analysis, they developed a concept termed "MileAge," a metabolomic age derived from these biomarkers. The term "MileAge delta" refers to the disparity between an individual’s calculated MileAge and their chronological age; a significant delta indicates the presence of accelerated aging.
The study found high performance consistency among the top-performing algorithms, which predominantly included tree-based ensembles and support vector regression models. Among the algorithms tested, Cubist rule-based regression proved to be the most effective, delivering results closely associated with established markers for aging and health. The researchers noted, “Across most models, individuals with an older metabolite-predicted than chronological age, indicating accelerated aging, were frailer, had shorter telomeres, were more likely to have a chronic illness, rated their health worse, and had a higher mortality risk.”
While the research demonstrated a clear connection between accelerated metabolomic aging and various health risks, it revealed that decelerated aging was not a consistent predictor of favourable health outcomes. Due to these findings, the researchers assert that metabolomics-based risk assessments should primarily serve to identify individuals at high risk of adverse health events.
The potential applications of the MileAge concept extend beyond the current study, as the researchers argue for further exploration of aging clocks based on different tissues and cells. They concluded, “Aging clocks hold substantial promise for research on life span and health span extension, as they provide an aging biomarker that is potentially modifiable.”
This pioneering research highlights both the progress made in AI technology's integration into healthcare and its potential to evolve in the future, providing powerful tools for understanding and monitoring human aging.
Source: Noah Wire Services