Biological Age: Evidence-Based Medicine or Marketing?
How old are you "really"? Longevity apps promise an answer in five minutes. Scientific papers are more cautious: they discuss mortality risk prediction, not a precise "wear-and-tear age." Between marketing and evidence-based medicine lies a gray zone — and that is where Health Vault works with biological age.
Three Generations of Aging Clocks
First Generation: Anthropometrics
BMI, blood pressure, VO₂max — "functional age" formulas have been popular since the 1980s and still appear in fitness apps. The problem is enormous variance. Blood pressure spikes after coffee; VO₂max depends on whether you ran for the bus an hour ago. Reproducibility is low; predictive power is lower still.
Second Generation: Epigenetic Clocks
In 2013, Steve Horvath showed that DNA methylation at 353 CpG sites predicts chronological age within ±3.6 years across 51 tissue types. GrimAge (Lu, Horvath et al., 2019) went further — predicting time-to-death from plasma proteins and methylation patterns. In cohort studies, GrimAge is significantly associated with all-cause mortality, coronary heart disease, and cancer.
Accurate and validated — but methylation analysis costs $300–500, requires a specialized lab, and does not scale for everyday health monitoring.
Third Generation: Phenotypic Age from Blood Chemistry
Morgan Levine (2018) proposed PhenoAge — a model using 9 markers from a standard blood panel plus chronological age: albumin, creatinine, glucose, CRP, lymphocyte percentage, MCV, RDW, alkaline phosphatase, and white blood cells. Trained on NHANES III (9,926 participants), validated on NHANES IV (11,432 participants, 1,012 deaths over 12.6 years). PhenoAge correlates with age (r = 0.94) but predicts mortality more accurately than age alone.
This is the only scalable approach — tests that millions of people already take every year.
When Biological Age "Lies"
Temporary distortions. A cold one week before testing: CRP and WBC spike — the model may "age" you by 10 years. Intense exercise: creatinine and ALP rise. Vegetarian diet: albumin below the lab reference without liver dysfunction. Menstrual cycle effects in women. A single measurement is often noise, not signal.
Population bias. PhenoAge was trained on NHANES (US population). Reference ranges for other populations may differ. Direct transfer without recalibration is an assumption to acknowledge.
Correlation is not causation. High phenotypic age may reflect chronic disease, stress, or an unlucky testing day. The model captures a pattern but does not explain the cause.
How Health Vault Uses Biological Age
We adopted a factor-centric model: every document is decomposed into atomic facts with LOINC codes. All 9 PhenoAge markers are in our database — the vector assembles automatically from uploaded labs.
The calculation is not limited to nine markers. We use all available biomarkers across dimensions: inflammation, metabolism, liver, kidneys, hematology, hormones. More data means higher model confidence.
The key advantage is dynamics. A single number — "you are 47 with a passport age of 42" — says little. A trend from +5 to +2 over two years is analytics. A rising trend is a reason to see a physician, not a longevity coach.
Can You Actually "Get Younger"?
Fitzgerald et al. (2021): RCT in 43 men aged 50–72, 8 weeks of diet, sleep, exercise, and relaxation. Epigenetic age decreased by 2.04 years vs +1.19 in controls (difference 3.23 years, p = 0.018). DO-HEALTH (2024): vitamin D, omega-3, and home exercise slowed PhenoAge acceleration in adults over 70.
Biological age is not a sentence — it is a parameter you can influence.
Practical Guidelines
- Do not panic over a single number — retest in 3–6 months.
- Watch the trend, not the absolute value.
- Control testing conditions: not after exercise, illness, or acute stress.
- This is not a diagnosis — discuss results with your physician.
The Health Index in Health Vault
Beyond biological age, Health Vault calculates a health index — an aggregated score across body systems. It combines biomarkers into categories: metabolic health, liver function, kidney function, hematology, inflammation, and hormonal balance. Each category receives a status (normal, attention needed, critical) based on the latest values and trends.
The index is designed for clarity, not alarm. A single elevated CRP after a cold should not dominate your overall score when other markers are stable. Weighting considers both current values and directional trends — improving glucose over six months matters even if the latest reading is still borderline.
Conclusion
Biological age is a monitoring tool under two conditions: structured data and longitudinal observation. Health Vault provides both through automated document processing and biomarker tracking over time. Used responsibly — as one input among many — it turns routine blood work into actionable insight.
Originally published on Habr