Biological Age Calculator (PhenoAge)
Your Phenotypic Age from your age and nine routine blood tests, with the formula from Levine et al. 2018, and which results push it older or younger.
Key facts
- Phenotypic Age combines calendar age with nine routine blood tests: albumin, creatinine, glucose, C-reactive protein, lymphocyte percent, mean cell volume, red cell distribution width, alkaline phosphatase and white cell count.
ExamplePhenotypic Age
42.2years
7.8 years younger than your calendar age of 50.
What moves it
◀ younger · older ▶
Years each result adds or takes off, compared with someone aged 50 with the average results of US adults in 1999 to 2010 (Phenotypic Age 44.0).
- CRP1 mg/L−1.5 years: C-reactive protein 1 mg/L takes off 1.5 years
- RDW13%+1.1 years: Red cell distribution width 13% adds 1.1 years
- Glucose90 mg/dL−0.9 years: Glucose 90 mg/dL takes off 0.9 years
- WBC6 10³/µL−0.5 years: White blood cells 6 10³/µL takes off 0.5 years
- Albumin4.4 g/dL−0.5 years: Albumin 4.4 g/dL takes off 0.5 years
- Creatinine0.9 mg/dL+0.5 years: Creatinine 0.9 mg/dL adds 0.5 years
- ALP70 U/L0.0 years: Alkaline phosphatase 70 U/L about no change
- Lymphocytes30%0.0 years: Lymphocytes 30% about no change
- MCV90 fL0.0 years: Mean cell volume 90 fL about no change
44.0 at average results − 1.7 from yours = 42.2
- Calendar age
- 50 years
- Average results at 50
- 44.0 years
- Yours vs average results
- −1.7 years
General information from published research, not a diagnosis or medical advice. Talk to a doctor before changing medication or treatment. Phenotypic Age is a research estimate from a population model. SourcesTerms
How it’s worked outLevine et al. 2018, with your numbers
Each result in the paper’s unit, times its weight from Levine et al. 2018 (Supplementary Table S1):
| Marker | Value | Weight | Product |
|---|---|---|---|
| Albumin g/L | 44 | -0.0336 | -1.4784 |
| Creatinine µmol/L | 79.56 | 0.0095 | 0.7558 |
| Glucose mmol/L | 5 | 0.1953 | 0.9756 |
| ALP U/L | 70 | 0.0019 | 0.1330 |
| WBC 10³/µL | 6 | 0.0554 | 0.3324 |
| Lymphocytes % | 30 | -0.012 | -0.3600 |
| MCV fL | 90 | 0.0268 | 2.4120 |
| RDW % | 13 | 0.3306 | 4.2978 |
| CRP ln mg/dL | ln(0.100) | 0.0954 | -0.2197 |
| Age | 50 | 0.0804 | 4.0200 |
| Constant | -19.9067 | ||
| Sum, xb | -9.0381 | ||
Mortality score M = 1 − exp(−exb × (e120 × 0.0076927 − 1) ÷ 0.0076927) = 0.0232
Phenotypic Age = 141.50225 + ln(−0.00553 × ln(1 − M)) ÷ 0.090165 = 42.22
M is the model’s 10-year risk of death for an adult in the 1988 to 1994 survey with these numbers: a step in the formula, not a forecast for you. The second line turns it into the age at which the average person in that survey had the same risk. Because the steps cancel, each unit of a marker moves Phenotypic Age by its weight ÷ 0.090165 years: glucose 2.17 years per mmol/L, RDW 3.67 per percentage point, albumin -0.37 per g/L.
What it can and can’t tell youBuilt on NHANES III, checked on NHANES IV
What it was built and tested on. 9,926 adults aged 20 and over in NHANES III (1988 to 1994), a national US survey, with over 23 years of death records; then 6,209 adults in NHANES IV (1999 onwards) with up to 12 years. There, Phenotypic Age correlated with calendar age at r = 0.94, and at the same calendar age each extra year went with a 9% higher risk of death from any cause, 10% from cardiovascular disease and 20% from diabetes.
A few years under is ordinary. The average results of US adults in 1999 to 2010 (11,432 people, Liu et al. 2018) give a Phenotypic Age of 40.0 at their average age of 45.5. So compare your answer with the “average results at your age” figure as well as with your calendar age.
What it can’t do. It was fitted on US adults aged 20 and over, with the lab methods of the time. One blood draw moves with recent illness, especially CRP, and with the lab. It doesn’t diagnose anything: a result flagged on your lab report is for your clinician, whatever this number says.
Units and conversionsUS and SI lab reports
- Albumin: g/L = g/dL × 10.
- Creatinine: µmol/L = mg/dL × 88.4 (creatinine weighs 113.12 g/mol).
- Glucose: mmol/L = mg/dL ÷ 18.016 (glucose weighs 180.16 g/mol).
- C-reactive protein: the formula uses the natural log of mg/dL, and mg/dL = mg/L ÷ 10. A result printed as “<0.3” counts as 0.3, so the answer is an upper limit; anything below 0.1 mg/L, including 0, counts as 0.1 mg/L.
- Alkaline phosphatase: U/L = µkat/L × 60.
- White cells: 10³/µL, 10⁹/L, K/µL and thousand/µL are the same number; cells per µL ÷ 1000.
- Lymphocytes: the formula uses the percent of white cells; a count in 10³/µL is turned into a percent with your white cell count.
- Mean cell volume in fL and red cell distribution width as the CV in % need no conversion. If your report gives RDW-SD in fL, it can’t be converted: use the RDW-CV.
How this works: Method, 3 sources, Checked against 5 worked examples,
How this works
Method
Phenotypic Age follows Levine et al. 2018: each blood result, in the paper’s units, is multiplied by its weight from Supplementary Table S1 and added to 0.0804 × age and the constant −19.9067; that sum gives a Gompertz 10-year mortality score M, and Phenotypic Age = 141.50225 + ln(−0.00553 × ln(1 − M)) ÷ 0.090165. Units are converted explicitly (CRP enters as the log of mg/dL), and each result’s bar is the years it adds or takes off compared with the NHANES IV average results (Liu et al. 2018).
Sources
- An epigenetic biomarker of aging for lifespan and healthspan (Levine et al.) (Aging (Albany NY), 2018)
- A new aging measure captures morbidity and mortality risk across diverse subpopulations from NHANES IV (Liu et al.) (PLOS Medicine, 2018)
- BioAge: biological age calculations using several biomarker algorithms (Kwon and Belsky) (R package)
How it’s tested
5 worked examples for this page are checked by automated tests before every release: given the inputs, the tool must show the expected answer.
Limits
General information from published research, not a diagnosis or medical advice. Talk to a doctor before changing medication or treatment. Phenotypic Age is a research estimate from a population model.
Privacy
On your device. Nothing you enter or open leaves this browser. About privacy labels
Changes
- First published.
Frequently Asked Questions
What is Phenotypic Age?
Phenotypic Age (PhenoAge) is an estimate of age from blood tests, published by Morgan Levine and colleagues in 2018. They took 42 clinical measures from 9,926 US adults in the NHANES III survey (1988 to 1994), with over 23 years of follow-up, and kept the nine that, together with calendar age, best predicted death from age-related disease. The result is expressed in years: the calendar age at which the average person in that survey had the same predicted risk.
Is this the same as an epigenetic or DNA methylation age test?
No. The same paper also built DNAm PhenoAge, an epigenetic clock that predicts Phenotypic Age from DNA methylation. Commercial biological age kits usually measure methylation. This calculator works out the blood-chemistry Phenotypic Age the clock was trained on, from results you may already have from a routine checkup.
Which blood tests do I need?
A comprehensive metabolic panel (albumin, creatinine, glucose, alkaline phosphatase), a complete blood count with differential (white cell count, lymphocyte percent, mean cell volume, red cell distribution width) and a C-reactive protein test, usually high-sensitivity CRP. Use results from the same blood draw where you can.
Why does another calculator give me a different number?
Usually units. The formula expects albumin in g/L, creatinine in µmol/L, glucose in mmol/L and CRP in mg/dL; a CRP in mg/L read as mg/dL adds about 2.4 years. Some later papers reprint the equation with typos (one prints 0.0195 for the glucose weight instead of 0.1953, which changes the answer by about 10 years), and questionnaire quizzes use a different idea altogether. This calculator uses the weights from the original paper’s Supplementary Table S1, converts every unit explicitly and matches the BioAge research package to within 0.1 year.
My CRP says “<0.3” or 0. What do I enter?
Type it as printed, “<0.3”. The calculator counts it as 0.3, so the answer is the most your Phenotypic Age could be: every halving of CRP takes about 0.7 years off. A CRP of 0 can’t go into the formula’s logarithm, so anything below 0.1 mg/L counts as 0.1 mg/L, and the page says so.
How accurate is it for one person?
It describes risk across a population, not a measurement of your body. A single draw moves with recent illness (CRP especially), lab differences and normal day-to-day variation. In the InCHIANTI study, people’s Phenotypic Age relative to their calendar age correlated only r = 0.50 between two visits nine years apart. Saving several draws and looking at the trend says more than one number.
Can Phenotypic Age say whether I am healthy?
No. It is a research estimate, not a diagnosis. Your lab report’s reference ranges, and a clinician who knows your history, are what to go by for any single result. The paper reports associations with mortality and disease, not what changes them, so this page makes no treatment or supplement suggestions.