(Image by frank333 on Shutterstock)
In a nutshell
- Japanese researchers developed an AI model that analyzes 22 steroid hormones in blood to determine your biological age—potentially revealing how fast you’re aging compared to your actual years.
- Cortisol, the body’s stress hormone, emerged as the strongest predictor of accelerated aging, with doubled levels associated with a 40% increase in biological age.
- The research found that biological age varies more widely as we get older, with sex-specific differences in hormone patterns—suggesting personalized approaches may be needed for addressing age-related conditions.
OSAKA, Japan — We all know people who seem younger—or older—than their actual age. Now science has a way to measure this difference. Researchers at Osaka University have found that the steroid hormones flowing through our bloodstream can reveal our true biological age, which often differs from the number of candles on our birthday cake.
Their study, published in Science Advances, uses artificial intelligence to analyze blood samples and determine how old your body actually is. This could change how doctors spot age-related diseases early and help us understand why some people age faster than others.
Counting our age in years since birth doesn’t tell the full story of how “old” we really are. Scientists have long suspected that our bodies age at different speeds. Two 50-year-olds might look and feel decades apart in terms of health, depending on their genes, lifestyle choices, and environment. The trick has been finding reliable markers that accurately capture these differences.
“Our bodies rely on hormones to maintain homeostasis, so we thought, why not use these as key indicators of aging?” says Dr. Qiuyi Wang, co-first author of the study, in a statement. Wang and colleagues discovered that stress hormones, especially cortisol, strongly correlate with aging. If you’ve ever felt like stress is aging you prematurely, this study suggests you might be right.
The study confirms something many doctors have observed—as we get older, the gap between chronological and biological age widens. The researchers liken this effect to “a river widening as it flows downstream.”
Simple blood test reveals biological age
Using just five drops of blood, the scientists collected samples from 148 people between ages 20 and 73. They measured 22 different steroid hormones in each sample using an advanced technique called liquid chromatography-tandem mass spectrometry. They built separate models for women and men after noticing important sex-based differences in hormone patterns.
The team developed a deep neural network model that incorporates steroid metabolism pathways, making it the first AI model to explicitly account for the interactions between different steroid molecules. “Our approach reduces the noise caused by individual steroid level differences and allows the model to focus on meaningful patterns,” explains Dr. Zi Wang, co-first and corresponding author of this work.
Cortisol was the strongest predictor of biological age for everyone. When levels doubled, predicted biological age jumped by approximately 1.5 times “Stress is often discussed in general terms, but our findings provide concrete evidence that it has a measurable impact on biological aging,” says Professor Toshifumi Takao, a corresponding author and an expert in analytical chemistry and mass spectrometry.
In women, several other hormones showed strong links to aging, including 17α-hydroxyprogesterone, cortisone, 11-deoxycortisol, and tetrahydrocortisol. For men, pregnenolone and testosterone were also key indicators of aging processes.
Hormones can help doctors tailor therapies for individuals
These findings might eventually lead to more personalized healthcare. By checking someone’s hormone profile, doctors could potentially spot accelerated aging before symptoms appear and recommend targeted interventions.
The study also looked at how lifestyle affects aging. Male smokers showed significantly faster aging than non-smokers. Oddly, female smokers didn’t show the same acceleration. The researchers think this might be because men in the study smoked more heavily, or because unmeasured factors in women (like diet or exercise habits) masked smoking’s effects.
Unlike previous research that treated aging variations as statistical noise, this study embraced this diversity as meaningful. Instead of looking at absolute steroid levels—which can vary widely between individuals—the model examines steroid ratios, providing a more personalized and accurate assessment of biological age.
Their approach doesn’t just spit out numbers—it helps explain the biochemical processes driving aging. By structuring their analysis to mirror actual hormone pathways in the body, the researchers created a model that offers insights into why certain hormones impact aging so strongly.
The study also helps explain why aging experiences vary so dramatically between individuals. By measuring how biological aging differs across populations and identifying key hormonal drivers, the research helps explain why some people seem to age more gracefully than others.
‘Just the beginning’ for aging research
This hormone-based approach complements other aging measures like epigenetic clocks based on DNA methylation patterns. Together, these techniques are building a more complete picture of how our bodies age at the molecular level.
As populations worldwide grow older, understanding the fundamentals of aging becomes increasingly important. “This is just the beginning,” says Dr. Z. Wang. “By expanding our dataset and incorporating additional biological markers, we hope to refine the model further and unlock deeper insights into the mechanisms of aging.”
Future studies might explore whether changing certain hormone levels through medication, lifestyle changes, or stress reduction could slow biological aging.
With ongoing advancements in AI and biomedical research, the dream of accurately measuring—and even slowing—biological aging is becoming increasingly feasible. For now, the research serves as a reminder that aging involves far more than birthdays—it’s a complex biological process shaped by our hormones and lifestyle choices, especially the stress we carry through life.
Paper Summary
Methodology
The research team took a clever approach to studying how hormones relate to aging. Using just five drops of blood from each participant, they collected samples from 148 people aged 20 to 73 and used a technique called liquid chromatography-tandem mass spectrometry to measure 22 different steroid hormones. They split their samples into two groups: 98 for building their model and 50 for testing it.
Before analysis, they had to solve a common problem in hormone research—levels vary naturally throughout the day and between testing batches. They developed a special scaling method that preserves the relative amounts of different hormones while removing these unwanted variations.
The heart of their approach was a neural network—a type of artificial intelligence—designed to mimic actual hormone pathways in the body. This wasn’t a generic black-box AI; it was carefully crafted to match real biology, starting with the hormone pregnenolone and following natural metabolic routes. They built separate models for women and men after noticing clear sex differences in hormone patterns. They also created a custom mathematical function that accounts for the fact that aging becomes more variable as we get older.
Results
The study uncovered several patterns linking hormones to biological aging. The standout finding was cortisol’s strong influence—when levels doubled, predicted biological age increased by approximately 1.5 times. This highlights stress hormones’ central role in aging.
Women and men showed different hormone patterns. In women, several hormones like 17α-hydroxyprogesterone, cortisone, 11-deoxycortisol, and tetrahydrocortisol were strongly associated with biological age. In men, pregnenolone and testosterone showed important relationships with aging.
When they examined lifestyle factors, they found that male smokers showed signs of accelerated aging compared to non-smokers, while female smokers didn’t show the same pattern. Their model successfully captured increased aging variability over time, confirming that as we age, biological age can diverge more widely from chronological age. Most predictions fell within what they called “physiological thresholds”—a twofold change from chronological age.
Limitations
The researchers openly acknowledge several shortcomings in their work. With only 148 participants, the sample size is relatively small, limiting how broadly their findings apply. Their validation group lacked detailed information about lifestyle factors beyond smoking—things like alcohol consumption, diet, and exercise patterns that likely affect biological aging.
Another limitation is their static measurement approach. Hormones fluctuate throughout the day and in response to various conditions, but the study took just one measurement per person. The researchers also note that their specialized neural network design makes direct comparisons with other aging prediction methods difficult.
For more definitive results, future studies should track people over time, include more diverse participants, and collect comprehensive lifestyle data to better understand how environment and behavior interact with hormones to influence aging.
Discussion and Takeaways
This hormone-focused research offers fresh perspectives on aging processes and potential applications. By showing that steroid profiles can reliably predict biological age, the study reveals how hormonal systems affect aging throughout the body.
The strong link between cortisol and faster aging underscores chronic stress’s potential long-term health impacts, suggesting that stress-reduction might be an important part of healthy aging strategies. The distinct hormone patterns in women and men point to the need for sex-specific approaches to age-related conditions.
The researchers believe this AI-powered biological age model could pave the way for more personalized health monitoring. Future applications may include early disease detection, customized wellness programs, and even lifestyle recommendations tailored to slow down aging.
The model’s ability to capture increased aging variability over time matches clinical observations that health trajectories become more diverse as people age. Looking ahead, this approach might eventually help identify people at higher risk for age-related diseases before symptoms appear.
Funding and Disclosures
This study received partial funding from the Japan Society for the Promotion of Science through a Grant-in-Aid for Specially Promoted Research (JP17H06096) awarded to Toshifumi Takao. The researchers declared no competing financial interests. They’ve made their data and analysis code publicly available through Code Ocean under appropriate licenses (CC-BY-NC for data and MIT license for code), promoting scientific transparency.
Publication Details
The paper “Biological age prediction using a DNN model based on pathways of steroidogenesis” appeared in Science Advances on March 14, 2025 (Vol. 11, Article ID: eadt2624). Qiuyi Wang, Zi Wang, Kenji Mizuguchi, and Toshifumi Takao from Osaka University’s Institute for Protein Research authored the study, with Wang and Wang noted as equal contributors. The research was submitted in November 2024, accepted in February 2025, and published that March with DOI: 10.1126/sciadv.adt2624.
