Modeled trajectory
One curve. Every assumption visible.
Expected age
—
— years remaining
Primary longevity model
Phenotypic age
—
9 biomarkers required
Complementary · zero model weight
Achievable goal
—
—
Middle 80% survival range
—
Reflects human variability, not full model uncertainty.
Model confidence
Developing
Complementary laboratory metric
Phenotypic Age
Zero model weightAn independent cross-check calculated from chronological age and nine blood biomarkers. PhenoAgeAccel removes the age pattern using a documented NHANES reference fit. Neither result changes the main expected-age estimate or factor attribution.
Laboratory age
—
Awaiting 2 biomarkers
PhenoAgeAccel
—
NHANES-adjusted
7/9 biomarkers available
Needs hs-CRP and lymphocyte %
Review laboratory inputs August 14, 2026 values prefilled
Where to find these results: most values come from two common lab reports. Your Comprehensive Metabolic Panel (CMP) usually contains albumin, creatinine, glucose and alkaline phosphatase. Your CBC with differential usually contains lymphocyte %, MCV, RDW and white blood cells. hs-CRP is usually ordered separately. Match the units shown below before entering a result.
All nine biomarkers are required; missing values are never imputed. PhenoAgeAccel is the laboratory age minus the age-specific value expected from Yirelo’s survey-weighted NHANES 1999–2010 reference fit. The raw age gap remains visible separately.
Survival curve
Probability of being alive
Current
Goal
Population
Against population baseline
Factor attribution
Attribution allocates the joint estimate; it is not proof that changing one factor causes the displayed number of years.
Decision support
What matters next
Hold every input constant
What if today’s profile stayed frozen?
Each future point advances your age while keeping every entered biomarker and lifestyle setting unchanged.
Time projection
Expected age if inputs do not change
Frozen current
Frozen goal
The line can rise as survival to an older age becomes known. It does not mean waiting improves health; it is conditional life expectancy at each future age.
One variable at a time
Impact of a favorable or adverse change
Each row changes only that measurement by 10%, holds all other inputs constant, and shows the modeled change in expected age.
Interpret carefully
A sensitivity test, not a forecast
Percent changes are mechanical what-if tests. “Favorable” is the better modeled result of moving the measurement up, down, or leaving it unchanged. Categorical inputs such as smoking, social connection and sleep apnea, and category-scored resistance training, are not percent-scaled. This is not a clinical target, a promise of added years, or a substitute for medical guidance.
Yirelo Plus · longitudinal tracking
See every metric—and what it meant at the time.
Save dated checkpoints as your measurements change. Yirelo preserves the expected-age result calculated at each point so you can compare your health trajectory without rewriting history.
Metric evolution
Follow one signal through time
Latest expected age
—
Add checkpoints to see a trend
Complete metric history
One horizontal view of every model input
Expected age is the estimate Yirelo produced from the saved model and plan on that date. Historical values are not automatically recalculated when the research model changes.
Open research record
Every input, assumption and source
This is the audit trail for Yirelo v0.6. It separates published findings from Yirelo’s prototype choices, shows the exact treatment used in code, and records why some plausible longevity factors remain outside the score.
Research links open directly in this tab for reliable access. Use your browser’s Back button to return to Yirelo.
Reading standard
Association is not causation
“High” means the relationship is consistently documented or the source is an official baseline. It does not mean changing that metric will cause the exact modeled gain. Grades describe evidence fit for this prototype—not treatment advice.
High large cohorts, meta-analysis, or validated framework
Moderate consistent evidence with calibration limits
Emerging limited endpoint, measurement, or subgroup evidence
Not scored no eligible mortality coefficient
Evidence registry
Model foundations
The core reports, cohorts, consensus statements and validation work that shape the model architecture.
September 2026 review
Six evidence decisions
Only factors with an interpretable human mortality association and a reproducible input receive a provisional coefficient.
FactorDecisionReason
Calorie restrictionNot scored
Two-year randomized data improve risk factors and one aging-pace measure, but do not provide a mortality coefficient.
Intermittent fastingNot scored
Trials primarily measure weight and metabolic outcomes; effects are broadly comparable with continuous restriction.
Resistance trainingAdded · provisional
Large cohorts report an all-cause mortality association, including analyses adjusted for aerobic activity.
Sleep durationAdded · provisional
Dose-response meta-analysis supports a U-shaped association; Yirelo attenuates it because measurement and reverse-causation risks are substantial.
Social connectionAdded · provisional
Multiple mortality meta-analyses support the association; Yirelo uses conservative self-reported categories because no single measure is universal.
Telomere lengthNot scored
Associations exist, but assay units are not portable and genetically longer telomeres involve cancer trade-offs.
Zero-weight research record
Promising, but not model-ready
These entries document the human evidence, the exact exclusion rule, and the sources needed for future review.
Metric-by-metric audit
What enters the model
Each record distinguishes the published endpoint from Yirelo’s exact prototype coefficient and its known limitations.
Derived outputs
How every displayed result is produced
These are computational definitions, including the experimental and confidence displays that are not clinical endpoints.
Yirelomins & supplements
Reviewed, but not added to the score
These records cover biomarkers and compounds considered during the review. None changes expected age in Yirelo v0.6.
Current rule: a supplement needs a reproducible measurement, a defensible range, and mortality-calibrated evidence before it can receive any weight. Biomarker association alone is not enough.
Known limitations
What this model cannot claim
- The coefficients are evidence-synthesis choices; they were not copied from one validated all-cause mortality equation.
- Cause-specific factors such as blood pressure, lipids, Lp(a), kidney function and CAC are attenuated before entering the all-cause curve.
- The combined model has not been externally calibrated, prospectively validated, or shown to estimate causal treatment effects.
- Unknown values are neutral by design. That avoids invented risk but can understate risk and lowers confidence.
- Waist and BMI are blended to reduce double counting; overlap among the remaining cardiometabolic factors is not fully eliminated.
- Resistance training overlaps with fitness, sleep duration overlaps with apnea and health status, and social connection overlaps with mental health and behavior. Conservative coefficients reduce—but do not eliminate—double counting and reverse causation.
- The baseline uses the 2023 U.S. sex-specific period life tables. Chilean, cohort-specific, pregnancy and validated menopause-stage calibration remain pending.