Can a simple blood test reveal how well we are aging? A team of researchers from the University of Vienna and Nankai University believes the answer is yes. By combining advanced metabolomics, machine learning, and a novel network modeling tool, they have identified key molecular processes that link physical activity with healthier aging. Their work highlights aspartate as a dominant biomarker of physical fitness and points to new opportunities for monitoring and supporting active aging.
It is well established that physical activity protects mobility and reduces the risk of chronic illness. Yet the molecular mechanisms behind these benefits have remained largely unclear. To address this gap, the team asked a straightforward question: Can we see the effects of an active lifestyle directly in the blood of older adults and identify the molecules that matter most?
Body Activity Index and machine learning models
The researchers first created a Body Activity Index (BAI) by combining data from walking distance, chair-rise, grip strength, and balance tests. This composite score provides a robust picture of endurance, strength, and coordination. In parallel, they developed a Metabolomics Index based on the blood concentrations of 35 small molecules. Analysis of 263 samples from older adults showed a striking correlation between the two indices (Pearson coefficient 0.85, p < 1 × 10⁻¹⁹). In other words: blood chemistry closely mirrors physical performance.
To go further, the team applied five different machine learning models, ranging from classical statistical methods to advanced deep learning networks. The best-performing models achieved more than 91% accuracy in distinguishing active from less-active participants. Across all approaches, eight metabolites consistently emerged as predictors of activity: aspartate, proline, fructose, malic acid, pyruvate, valine, citrate, and ornithine. Among these, aspartate stood out as the strongest marker. Its predictive value was two to three times higher than the others. The study was published in npj Systems Biology and Applications.
Mapping the metabolic network with AI
Correlation alone does not explain how these molecules interact. To uncover deeper insights, the team used COVRECON, a modeling tool that reconstructs biochemical networks by analyzing how metabolites fluctuate together. This revealed two enzymes—aspartate aminotransferase (AST) and alanine aminotransferase (ALT)—as central hubs. Both are widely known from standard liver tests, but here they emerged as key indicators of how physical activity reshapes metabolism.
The prediction was confirmed in practice: over six months, AST and ALT fluctuated more strongly in active participants, pointing to greater metabolic flexibility in liver and muscle.
Links to brain health
The implications extend beyond muscle and mobility. Aspartate is also a neurotransmitter precursor in the brain, activating NMDA receptors critical for learning and memory. Low AST and ALT levels in midlife, or an elevated AST/ALT ratio, have previously been linked to higher risk of Alzheimer’s disease and cognitive decline.
This study therefore suggests a potential molecular bridge between physical activity and brain resilience. Regular exercise may not only preserve strength but also protect the brain, with changes in aspartate metabolism serving as measurable signals of benefit.
Toward precision health
For healthcare and research, the study offers a powerful message: physical activity rewires metabolism at the molecular level, and these changes can be tracked through simple blood tests combined with AI-driven analysis. According to lead researcher Wolfram Weckwerth, “Physical activity does more than build muscle mass, it reshapes our metabolism. By decoding those changes, we can track, and even guide, how well someone is aging.”
This integrative approach could form the basis for next-generation diagnostic tools, enabling clinicians to measure the impact of lifestyle interventions in real time and tailor recommendations for older adults. In the future, routine blood analyses enhanced with AI could become part of personalized strategies for healthy aging and dementia prevention.
AI-powered digital twin
Two weeks ago we wrote about the development of an AI-powered digital twin that predicts future disease risks with unprecedented precision. Using data from more than 30,000 volunteers and over 260 billion data points collected through the Human Phenotype Project, the model analyzes 17 body systems to forecast conditions years before they manifest.
This approach enables patients and physicians to simulate lifestyle and treatment scenarios. For example, testing whether dietary changes could cut diabetes risk in half. One breakthrough is the ability to calculate the biological age of individual organs. Deviations from expected aging patterns can flag early signs of disease, such as prediabetes, long before conventional tests. The data also revealed critical gender differences, including accelerated biological aging in women around menopause.
