Methodology

How BackNine scores you.

Every number in BackNine has an “Under the hood” view in the app that lists its inputs and how each one moved the result. This page is the same math, written out once, in public — so you can check it before you trust it. If we can't show the math, we don't show the score.

Biological Age

Your body's age as read from up to thirteen markers — wearable-derived (HRV, resting heart rate, VO₂ max, sleep, body fat), blood pressure, and labs you upload (HbA1c, LDL, HDL, triglycerides, fasting glucose, hsCRP, eGFR). Four steps:

1. For each marker, we compute the population-typicalvalue for your age and sex — what an average person your age actually measures, not an athlete's optimum. (Typical body fat for a 58-year-old man is about 26%, not 15%.)
2. Your reading becomes a z-score: how many standard deviations you sit from typical, in the direction that matters (lower LDL is better; higher VO₂ max is better).
3. Each marker's z-score converts to years using its clinical signal weight — how strongly that marker tracks with mortality in the literature — and is capped at ±6 years so no single lab result can dominate.
4. The weighted average of those year-deltas is added to your chronological age. We need at least three markers to show a number; fewer would be false precision.

Confidence is labeled high, medium, or low based on how many markers are present. When you have both creatinine-based and cystatin-C-based eGFR, we use cystatin C — it isn't skewed by muscle mass.

Weekly Health Span Score

The process score: what your habits did this week, from sensors only. Four bands, each with a point ceiling, normalized to 100 across whichever bands your devices can measure (minimum two).

• Sleep hours (15 pts) — 7-day average. Full marks at 7–9 h; 6.5–7 or 9–10 h scores lower; under 6 h scores lowest.
• Sleep consistency (10 pts, Oura) — bedtime standard deviation. Under 30 min = 10, under 45 = 7, under 60 = 4.
• Steps (10 pts) — 7-day average. 7,500+ = 10, 6,000+ = 8, 4,500+ = 6, 3,000+ = 4, else 2.
• Active days (15 pts) — days with ≥300 active kcal or a detected workout. 5+ days = 15, 4 = 12, 3 = 9, 2 = 6, 1 = 3.

Grades: 85+ Excellent, 70+ Good, 55+ Fair. This is your gross — the number your doctor cares about.

The Weekly League plays net, like golf

Four friends — one fit, one carrying extra weight, one much younger, one much older — can't compete on gross scores: the same person wins every week and the same person buys the coffee. So the League does what golf did a century ago. Everyone plays against par, and everyone's par is their own.

• Par — the average of your best 40% of daily Health Span scores in the 28 days before Monday (golf uses best 8 of 20). Dropping the bad days means a lazy month can't earn you strokes.
• Scratch — par is capped at 90. The score tops out at 100, so a very fit player's best days average 99 and he could never beat par. Above 90, the game is holding it.
• Slow up, fast down — par can fall at most 3 points from one week to the next, so a bad patch can't quietly hand you strokes; it rises as fast as you do.
• Net — this week's score minus par. +4 means you beat the person you've been by four. The League ranks on net; lowest net buys the coffee.
• New members — your watch or ring has been keeping score for months, so we read the 28 days before you joined and hand you a real par on day one.
• No device at all — your iPhone counts steps on its own. Connect Apple Health, log your sleep in ten seconds a day, and a 7,500-step day counts as active. Three days to a provisional par, ten to a full one — and while your history is short, par is your median day, not your best, so a strong first week can't set you up to fail.
• Medical hold — injured or sick? Set a date in your profile and you sit out until then. Golf has one too.

Fewer than ten days of history and your par is marked provisional (≈). Tap any player on the League to see their par, their net, and the bands behind the number.

QYL Index — Quality Years Left

The horizon number. Standard actuarial life tables give the expected remaining years for a person of your age and sex. We scale that by the share of remaining life the average US adult spends active and independent (about 70%) — that's healthspan, not lifespan. Then the key move: we evaluate those tables at your biological age, not your birthday age. Two years biologically younger means the table pays out as if you were two years younger.

BackNine-computed rings (Apple Health users)

Oura users see Oura's own Readiness, Sleep, and Activity scores. Apple Health users get scores we compute from the same kinds of inputs, marked “≈ computed by BackNine” so nobody mistakes them for a manufacturer's number.

• Sleep — duration band (85–100 for 7–9 h, tapering outside) × quality, where quality is the deep + REM share of the night (healthy ≈ 35–45%). Without stage data the score caps at 90: we won't certify a perfect night we didn't see.
• Activity — 70% from steps against 10,000, 30% from active calories against 600.
• Readiness — 75 baseline, moved up to ±60 points by how today's HRV and resting heart rate compare to your own 14-day baseline, clamped 35–98.

What we don't do

No score is adjusted for engagement, streaks, or how often you open the app. Family history is shown to you and included in doctor reports as screening context, but it is deliberately kept out of the Bio Age and QYL math — your measured markers already express your genetics, and parental lifespan is a weak, confounded signal. Nothing here is a diagnosis or a prediction for you as an individual; it's population statistics applied honestly, meant to be read with your doctor.