Evidence summary: EpiMedTech epiAge

epiAge is a saliva test sold by EpiMedTech Global, a company based in Singapore [8]. The company says HKG epiTherapeutics, a laboratory company in Hong Kong, has been its subsidiary since April 2024 [8], and its report template names HKG epiTherapeutics in Hong Kong as the testing laboratory [4]. You collect a saliva sample at home and post it back to the laboratory [12]. Results come through a phone app [3]. The company's pages give different minimum ages: 18 in some places and 21 in others [4][7][9].

On 3 November 2026, the company's website showed a price for epiAge but no way to order it online, and its product list showed "Contact for availability" [3][10]. Its Canadian arm says it is not taking new orders and has paused testing [3][11]. In Europe, distributors sell the test, including a German company that sells it under the epiAge name [12][14].

The company's sample report, dated February 2024, shows two results: "Your Epigenetic Age" and a box labelled "Technical Accuracy", which the report does not explain [4]. The report does not ask for your calendar age. Instead, it shows a chart of other customers' results and asks you to mark your own place on it [4]. The product page says the app can draw this chart for you if you enter your calendar age [3]. A 2021 sample result from a German distributor looks different. It gives a "biological age" to two decimal places, and it explains its "technical accuracy" percentage as how close the three runs of your sample were to each other [13]. We do not know what customers receive today.

DNA methylation is a chemical mark on DNA; at some sites it changes with age. It is read at CpG sites, places in the DNA where it can occur. An epigenetic clock is a formula that turns methylation at chosen CpG sites into an age-like number.

The laboratory extracts DNA from your saliva and reads methylation at a small number of sites in one short stretch of DNA, using a chemical treatment followed by DNA sequencing, which reads the DNA letter by letter [4]. The developers' paper and patent place this stretch in or next to a gene called ELOVL2 and count 13 sites in it [15][18]. That is the only measurement. The age is calculated by putting the readings into an equation [4]. Saliva contains a mix of cells from the lining of the mouth and white blood cells, and the developers note that the mix differs between children and adults [15].

Three different formulas have been described for this stretch of DNA. Keeping them apart matters for what the evidence shows:

• A published three-site formula, called EpiAgePublic. The developers published it in 2025 [15] and corrected a sign error in it later that year [16]. It uses 3 of the 13 sites, because those 3 can also be read on standard methylation chips, and it was built on chip data; most of its published checks also use chip data rather than the sequencing test [15].
• The formula behind the test you buy. The developers' paper says the company's commercial service "utilizes a proprietary model" for the same stretch of DNA, one "originally developed using saliva samples" [15]. In the sources we reviewed, this formula is not published.
• A formula in the developers' patent. The patent, first filed in 2019 and granted in the US in July 2026, sets out a 13-site equation for saliva [18]. The company says the current test is a "second generation" built on "more than 600 samples" [5]. In the sources we reviewed, we could not tell whether the patent equation is the one used today; the patent describes it as the equation for customers' saliva samples at the time of filing [18].

EpiMedTech epiAge is currently rated Insufficient on Compare [1]. The rating concerns the epigenetic age the test gives from saliva, worked out with the company's own formula, as an estimate trained to match calendar age. It does not cover the published three-site formula when researchers use it on its own, disease risk or diagnosis, or the lifestyle suggestions in the report. This page explains that rating. It does not change it.

We rate epiAge Insufficient for its own epigenetic-age result: in the sources we reviewed, no peer-reviewed study reports how accurate the formula behind the test is in saliva, the sample the test uses.

• The published formula is a different one. The three-site formula was built on chip data from the blood of 4,625 people [15]. On those same people, its estimates matched calendar age with a correlation of 0.87 [15]. A correlation measures how closely two numbers rise and fall together; 1 would be a perfect match. In chip data from the saliva of 609 people aged 9 to 91, the correlation was 0.93; one established clock reached 0.99 on the same data [15]. Both data sets span very wide age ranges (birth to 103 years in blood, 9 to 91 in saliva), and a wide range makes a high correlation easier to reach. The paper does not report an average error in years [15]. These figures describe the three-site formula on chip data. Our methodology rates a commercial test "on the evidence for that product’s own output, not for the scientific method behind it" [2].
• The test's own formula has been reported only in blood. The paper reports the commercial formula only in DNA from the blood cells of 54 people, average age about 72 [15]. There, its estimates matched calendar age with a correlation of 0.53 [15]. In a group this close in age, a high correlation is harder to reach, so this figure says little either way. That measurement was made by HKG epiTherapeutics' commercial service and paid for by HKG epiTherapeutics [15].
• Other figures are not peer-reviewed. The patent reports that a 13-site formula fitted on saliva from 65 volunteers matched calendar age in those same people with a correlation of 0.95; it does not say whether the equation it prints is that formula [18]. The patent does not describe testing either on other people, and a patent is not a peer-reviewed study. The company's report says the correlation with calendar age is "close to 0.8" [4], the company says the current version was developed on "more than 600 samples" [5], and the 2021 distributor sample says the test is accurate to "+/- 2.8 years" [13]. None of these comes with a published dataset.
• No independent check. In the sources we reviewed, no researchers without ties to the company have tested epiAge. A 2026 review by researchers outside the company says there is "No independent peer-reviewed validation of the commercial test" and that accuracy data "are not publicly available" [17]. The review's authors have founded, own shares in or worked for biotechnology companies [17]. The same review describes the test as reading 300 sites, which does not match the developers' own descriptions [15][17][18].

In the sources we reviewed, five criteria in our Insufficient tier fit: "Vendor-only claims without published validation", "No peer-reviewed studies or only vendor-funded studies without independent oversight", "Proprietary models without methodological transparency", "No independent replication or external validation" and "Marketing language exceeds published evidence" [2]. The second fits because the only peer-reviewed data for the test's own formula were measured by HKG epiTherapeutics' service, paid for by HKG epiTherapeutics and analysed by a company scientist; the academic co-authors recruited the participants and prepared the samples [15]. The third fits largely rather than fully, because the patent sets out a 13-site equation, though we could not tell whether it is the one used today [18]. One of the tier's examples, "Clocks comparing across different tissue types without calibration" [2], is close to this case: the formula is said to be built on saliva, but its only peer-reviewed results are in blood [15].

We considered the next tier up, Emerging, which describes "pilot human studies" [2]. We rejected it because the one peer-reviewed pilot of the test's own formula is in blood, not saliva, and was measured and paid for by the company's own laboratory. Our confidence in Insufficient is moderate. A peer-reviewed report of the current formula's accuracy in saliva, even one by the developers, could move the rating to Emerging. Reaching Promising would also need the formula to be published and either checked by researchers without ties to the company or validated in a large group of people [2].

• "Fully disclosed". A January 2025 press release says "For the first time, EpiAge’s algorithm and validation details are fully disclosed" [6]. The paper it refers to says the commercial service "utilizes a proprietary model" [15], and the same press release calls it "the proprietary EpiAge algorithm" [6].
• "Over 4,600 participants". The same release says the study "validates the algorithm through data from over 4,600 participants" [6]. Those 4,625 people are the blood chip data used to build the three-site formula, and its 0.87 figure was measured on the same people [15]. They do not test the formula used in the saliva test.
• "Clinical-grade". The product page calls epiAge "the only clinical-grade epigenetic test available today" [3]. In the sources we reviewed, the company does not define the term beyond running each sample three times [3], and the company's disclaimer says its products "are not intended as medical advice or diagnostic tools" [7].
• Disease and lifespan. The product page promises to help you "Foresee and prevent diseases before they manifest" and "Extend your lifespan" [3]. In the sources we reviewed, no study links an epiAge result to later disease or survival. The company's own report says "there is no clearly proven immediate correlation between older "epigenetic" age and chronic diseases" [4].
• FDA registration. In November 2024, the company announced that epiAge was "the first epigenetic age test in the world to be registered by the FDA" and said that "FDA registration ensures that the test complies with applicable standards for safety, quality, and reliability" [5]. FDA registration and listing is not approval or clearance. A US federal rule says registration "does not in any way denote approval of the establishment or its products" [19], and the FDA says a listing entry "does not denote approval, clearance or authorization" [20]. The company's own disclaimer says the service "has not been cleared or approved by the U.S. Food and Drug Administration" [7]. In FDA's public registration and listing data, as updated on 26 October 2026, we found no listing for epiAge, EpiMedTech or HKG epiTherapeutics, either by name or among the entries for the device category the company's announcement cited [5][21]. Those data show current entries only, so we cannot tell whether a listing existed earlier.

The product page says epiAge can help you "Decelerate the aging process" [3]. The company's own report is more cautious: "it is still not clear how one can decelerate his “epigenetic clock”", and it describes the effect of diet as "still an open question" [4]. In the sources we reviewed, no published study has tested whether any lifestyle change, supplement or treatment changes epiAge results from saliva. This page is not a recommendation to make any particular change.

The company runs each sample three times and says the spread of those three runs "averages only one year" [3][5]. It says this lets customers "distinguish between true biological effects from interventions and technical noise" [5], and the 2021 distributor sample says "the smallest changes can thus be detected" in a follow-up test [13]. Re-running one sample measures only variation in the laboratory. It does not show how much results vary between two saliva samples taken on different days. In the sources we reviewed, there are no such data for epiAge and no threshold for a meaningful change.

In the sources we reviewed, the only published repeat data come from the developers' paper. They re-measured blood DNA from 41 people four times each, using the three-site formula [15]. The uncertainty range around each person's average was under 1 year for 14 people and more than 2 years for 5 [15]. The developers use these data to argue for repeat runs: without them, they write, "it becomes challenging to discern whether observed changes in epigenetic age across multiple time points are due to actual biological effects, batch effects, or merely technical noise" [15]. Repeat runs deal with noise in the laboratory. They do not deal with differences between two saliva samples.

An epiAge result is one estimate of epigenetic age from saliva, made by the company's own formula, which is built to match calendar age [15]. A result younger or older than your calendar age is a gap at one point in time, not a speed of aging. The report's chart compares you with the company's customers who chose to share their age; the report does not say how many there are [4]. The report treats a result outside its 95% band as a "red flag" [4]. In the sources we reviewed, no study shows that such a result signals a health problem, and in the one study of the test's own formula, the gap between the result and calendar age did not differ clearly between people with Alzheimer's disease and people without it [15]. The company's report says the test "is not intended to be health information or medical data or to be used to screen, diagnose, treat, prevent or assess the risk of any disease or condition" [4]. The reviewed sources do not show whether using this test to guide decisions improves health.

All the accuracy data we found for epiAge's methods come from its developers or sellers. The first author of the 2025 paper is named as an inventor on the patent application that covers "the tests disclosed in the paper", and the last author is a shareholder in EpiMedTech Global [15]. The company names them as its Chief Innovation Officer and its Founder and CEO [6]. Both are named as inventors on the patent, which was granted in the US in July 2026 and is listed as assigned to EpiMedTech Global [18]. The paper was funded in part by EpiMedTech Global, and the measurement with the test's own formula was paid for by HKG epiTherapeutics [15]. The patent and the report template are company documents, not peer-reviewed studies [4][18]. The one outside review we found was written by researchers with ties to biotechnology companies [17].

In 2024 the company said samples are analysed in laboratories accredited by CAP (College of American Pathologists) and CLIA, the US laboratory certification scheme [5]. US government (CMS) records list a CLIA certificate for HKG epiTherapeutics' Hong Kong laboratory, certified in May 2023 and marked as CAP-accredited [22][23]. The same records list that certificate as terminated on 17 March 2026, with the reason code "voluntary – merger, closure", which does not say which of the two applied [22][23]. The sources we reviewed do not say which laboratory processes samples now. A European distributor's page says both that HKG epiTherapeutics' laboratory analyses the samples and that they are analysed in the research laboratory of Moshe Szyf of McGill University [12].

For current price and availability, see Compare. If you spot an error, please tell us.