Japanese researchers have developed a new, non-invasive method for measuring biological aging based on urine. By combining machine learning with microRNA analysis, this so-called “biological aging clock” is able to predict a person's age with a high degree of accuracy. The average deviation from chronological age is only 4.4 years.
The technology was developed by Craif Inc. in collaboration with Nagoya University's Institute of Innovation for Future Society. The results have been published in the scientific journal npj Aging and mark an important step towards accessible, preventive healthcare.
Chronological versus biological age
Age is one of the most important risk factors for chronic conditions such as cardiovascular disease, diabetes, and cancer. However, the number of years a person lives does not always reflect their actual state of health. Some people age faster than average, while others age more slowly. The difference between chronological and biological age can therefore provide valuable information for prevention and early detection.
Biological aging clocks attempt to clarify this difference using biomarkers that change with age. Until now, DNA methylation models have been the most widely known, showing strong links to disease and mortality risk. In addition, microRNAs (miRNAs) are gaining interest. These small RNA molecules play a role in the regulation of gene expression and are involved in age-related processes.
An aging clock based on urine
What distinguishes this study is the use of urine as a source. Urine is easy, painless, and inexpensive to collect, which makes large-scale application attractive. The researchers developed an aging clock based on miRNAs in extracellular vesicles in urine, using machine learning.
The study used urine samples from 6,331 people who participated in the miSignal Scan cancer screening test. In addition to the samples, questionnaire data was also collected on age, gender, body weight, smoking behavior, alcohol consumption, exercise, and comorbidity, among other things. Participants could opt out of the use of their data through an opt-out procedure.
The machine learning model was trained using data from 2,400 participants. It was then validated in two separate test groups: one with 2,840 individuals with a balanced distribution of age and gender, and a second independent group of 1,091 individuals without this balance.
High accuracy, broad applicability
After sequencing the urine samples, averaging approximately 4 million ‘reads’ per sample, 407 relevant miRNA characteristics remained for model development. In the training set, the model predicted age with an average deviation of 5.1 years. In the first test set, that deviation decreased to 4.5 years and in the second test set to 4.4 years.
It is striking that the model identified twenty miRNAs that change consistently with age. Ten of these increase with age in both men and women, four only in men, while six actually decrease. These miRNAs were found to be involved in known ageing processes, such as bone remodelling, immune function, mitochondrial dysfunction and apoptosis (programmed cell death).
Comparison with existing methods
In terms of accuracy, the urine-based aging clock lags slightly behind DNA methylation clocks, which are currently considered the ‘gold standard’. At the same time, the new method performs better than existing aging clocks based on miRNAs or mRNAs from blood. This makes the urine approach interesting, especially given its ease of use and low burden on the patient.
However, the researchers do have some reservations about its application at the extremes of the age spectrum. Accuracy decreases in individuals younger than 25 or older than 80, so caution is advised when interpreting the results.
Digital biomarker for preventive care
According to the authors, this is the first validated biological aging clock based on urine miRNAs that is both accurate and practical. The technology thus opens the door to new forms of preventive care, in which biological aging and accelerated aging processes can be detected at an early stage.
In the context of digital health and personalized care, this development fits well with the shift from “disease care” to “health care.” Non-invasive biomarkers, combined with AI, can support healthcare providers in risk stratification, lifestyle advice, and monitoring of interventions. Urine as a data carrier could well play a key role in this.
Urine toilet tester
A year ago aChinese startup, Shanmu, introduced S1 at CES in Las Vegas, a compact sensor-equipped device that turns any toilet into a home urine-testing system. Inserted into a toilet bowl, the iPhone-sized gadget analyzes just 1 microliter of urine during normal use and delivers results within 10 minutes via a mobile app.
The S1 can detect markers related to kidney and liver function, metabolism, hydration, acid–base balance, and inflammation. The app supports test histories for multiple users. The waterproof device runs one to two months on a charge, while sensors last about six months. Pricing details have not yet been disclosed.
