Life Expectancy Calculator.

We Personalize Everything to You

Every projection, every hour, every data point you see in Bank of Health is built on your actual health history and actual choices. Always.

This means that our "best case scenario" isn't some generic guideline - it's where you in particular could get, starting today, if you started doing all that you could for your long-term health.

The gains and losses we show you - both for today and long-term - are real, modeled against the best realistic version of your health. Not some hypothetical runner who only eats broccoli. Not a version of you who was never obese or never smoked or never had that rough patch in their 30's. You.

Everything we show you is actually on the table today. What you do with that is up to you.

We Build on Data, not Ideologies

A lot of people have opinions on the best way to live a long, healthy life. That's great for them. But here's the thing - there is actually a correct answer, as a species we already know it.

We've done population-level studies on millions of people to understand what helps and what hurts - and exactly how much.

That's the foundation that we've built Bank of Health's models on. Every data point, model, and projection we have is based exclusively on peer-reviewed, large studies.

The honest truth is that big-picture research on health is settled, and has been for a while. The studies we used are linked at the bottom - but, spoiler, in study after study we find the same thing about being healthy, and you already know what it is:

• Move an hour a day.
• Lift heavy things a couple times a week.
• Keep a healthy weight.
• Maintain active, real social connections you care about.
• Don't smoke.
• Don't drink much or at all.

That's it.

Do those things - in whatever way works for you - and you've given yourself the best chance to live to a healthy, active, ripe old age.

It's your life, live it how you want.

Just like a bank won't tell you how to spend your money, we're not here to tell you what to do with the time you have.

Want to have those extra drinks? Go for it. Want to keep smoking, sit 14 hours a day, or skip leg day? All good.

We just believe you deserve to know, as accurately as possible, what each one of those choices does.

So that's what Bank of Health does. We help you track what your choices do, and make informed decisions you're happy with.

And that's all we'll ever do.

No sneaky sponsored links, no pitches to a workout plan, no hawking green hipster juice for $20 a bottle, no ads, no selling your data or skewing our models.

Real models based on real data, from real, highly-reviewed research, made simple, and consistent respect for your autonomy.

How We Calculate Your Daily Numbers

Each day you log data, Bank of Health:

1. Aggregates your inputs - workout MET-minutes (from your wearable, manual entry, or other sources), sitting band, sleep hours, alcohol, smoking, weight, and social engagement.
2. Maps inputs to factor HRs - activity uses a 7-day rolling average of weekly MET-minutes; strength uses calendar-week session count (resetting on your chosen day); sedentary uses today's sitting band with a PA offset from today's total MET-minutes; sleep, drinking, and social use today's values; BMI uses current weight + your profile height, mapped through your region's BMI–mortality curve.
3. Computes combined HR - multiplies all eight factor HRs together.
4. Calculates net hours - compares your combined HR to the population baseline (1.0) and converts the difference to hours of healthy life expectancy gained or lost that day. The conversion uses your country- and sex-specific life tables to determine your remaining life expectancy, then prorates the annual HR effect to a single day.

These daily hours accumulate in your ledger. On a good day (combined HR < 1.0), you earn hours. On a bad day (combined HR > 1.0), you spend them. Your lifetime total reflects the cumulative effect of your daily choices.

How We Calculate Your Life Expectancy

Who's Behind This?

Me. Hi, I'm Steven Skoczen, the guy who built Bank of Health.

I'm a Cornell-trained engineer, indie software developer, and co-author of peer-reviewed research in JAMIA, Annals of Behavioral Medicine, and The Journal of Rheumatology. I've been helping scientists build software that matches their research since 2000.

I built Bank of Health to scratch my own itch - a tool to help me make daily life choices that was rooted in the research, but still easy to use. I continue to improve and maintain the tool, including updating the model when new research is available. Our methodology and this page were last reviewed July 2026.

The Eight Factors

Each factor has specific hazard ratios derived from large-scale epidemiological studies. Here are the exact values we use:

Between the discrete bands, we use smoothed interpolation - piecewise linear curves anchored at the literature thresholds. This means 300 MET-min/wk of activity gets a proportional benefit between the 0 and 420 MET-min/wk anchors, rather than jumping between bands.

Recovery Curves

When you change a behavior, the health benefit (or harm) doesn't appear instantly. Past behavior leaves a residual effect that decays exponentially over time. We model this with factor-specific half-lives:

This means your decade history matters. If you smoked for 20 years and quit last year, your effective smoking HR is still elevated - it will take several years to approach a never-smoker's risk. The model captures this transition realistically rather than treating a lifestyle change as an instant reset.

• Smoking decays from your current-smoker risk toward a residual floor over a few years. How high that floor sits depends on how long you smoked: a few years of smoking recovers essentially all the way back to a never-smoker (floor ~1.0), while a lifetime of smoking leaves a permanent floor as high as ~1.7 from accumulated damage (Jha 2013). Most of the excess above that floor can still be won back.
• Weight works in two parts. First, how long you carried excess weight sets the size of the risk - a year of obesity barely registers, but many years compound into a larger hazard (see Duration Amplification below). Second, that burden is not locked in: the duration amplifier only applies while you're actually obese, and the obese-years it counts wind back down over your lean years - so getting back to a healthy weight brings you toward the mean rather than leaving you stuck. The BMI hazard itself then decays back toward healthy over a couple of years, toward a residual floor that depends on how long you carried the weight: a year or two of obesity leaves essentially no permanent mark (~0% residual), while decades leave a bounded scar of up to ~35% of the excess - even maximal sustained weight loss (bariatric surgery) doesn't fully erase a lifetime of obesity (SOS trial; Framingham metabolic scarring). Everything above that floor is still yours to win back, and for most people who've recently gained weight, that's nearly all of it.

Healthy Years vs. Impaired Years

Not all years of life are equal. Some you're independent, healthy, and active. Some, if you live long enough, will be impaired - you'll need help from other people for basic daily tasks.

But there's a key insight from the research: exercise and strength training don't just add years to your life - they dramatically compress the period of disability at the end. The impaired period shrinks, meaning more of your remaining life is spent healthy and independent.

So, our model distinguishes between healthy years (independent, functional living) and impaired years (significant disability requiring assistance), using the Sullivan method: for each age we hold a baseline fraction of survivors who are living independently, and weight your survival curve by it.

Each factor affects disability differently than mortality. We use disability shift multipliers (DSM) to capture this:

How the shift becomes years:

At each age we convert your hazard ratios into a "disability age" offset:

disability shift = Σ log₂(factor HR) × 8 × DSM

And then we add that disability shift to your age. The log₂ × 8 turns a hazard ratio into an approximate "years of aging" equivalent (8 years per doubling of mortality risk), and the DSM scales it by how much that factor drives disability beyond death. We then read the healthy fraction at your effective age rather than your calendar age. Protective factors (exercise, strength) produce a negative shift - you're "disability-younger" than your years; harmful ones push the lookup older.

Weight is handled differently. For weight, disability and mortality aren't proportional - obesity raises disability far more than it raises the risk of dying - so weight's disability effect isn't read from its mortality hazard at all. We model it straight from the Healthy→Disabled transition rates in Reuser et al 2009 (sex-split, because the effect differs markedly between men and women), calibrated to reproduce their measured loss of disability-free years and held flat above the top of their data (BMI 35+). Weight also still shortens life through the mortality side (using the larger Di Angelantonio data) - so for a heavier person you'll see both effects: fewer total years, and a larger share of the remaining ones impaired.

Smoking is handled the same way - in mirror image. Smoking is the opposite of weight: it's mortality-dominant ("smoking kills"), so its disability effect is small. We read it per-state directly from Reuser's Healthy→Disabled rates - a current smoker carries a real but modest disability shift; a former smoker sits back near baseline - independently corroborated by Klijs et al 2011, which found smokers spend essentially the same number of years disabled as non-smokers. When you quit, that disability shift eases smoothly down to the ex-smoker level as your body recovers, over the same few-year timescale your mortality risk takes to fall. The heavy, lasting cost of smoking is on the mortality side, which is unchanged.

Duration amplification does not feed this shift. The obesity duration multiplier below comes from a mortality study and scales your mortality risk only. The disability multipliers come from studies that measured baseline status, not exposure duration, so the shift uses your un-amplified hazard ratios. This keeps the two effects from being double-counted.

Duration Amplification

For obesity, how long you've been exposed matters independently of your current state, so we apply a duration amplifier based on cumulative years of exposure:

Obesity: Based on the Framingham Heart Study (Abdullah et al., 2011), which measured all-cause mortality by cumulative years lived with obesity: HR 1.51 at 1–5 years, 1.94 at 5–15, 2.25 at 15–25, 2.52 at 25+. The cross-sectional BMI hazard above already reflects a typical obese person — who in these cohorts had been obese about 8 years — so rather than stacking the full duration curve on top (which would double-count), we re-center it there: the amplifier is 1.0 at ~8 years, dips to about 0.82 for someone newly obese, and rises to about 1.37 for lifelong obesity. And because losing the weight counts, your accumulated obese-years wind back down during lean years rather than being locked in. It comes from mortality research, so it scales mortality risk only - not the disability shift described above.

Smoking has no separate duration amplifier. Instead, smoking's exposure-duration effect is built directly into the curves above: the current-smoker hazard ratio varies by attained age (drawn from cohorts of typical, lifelong smokers, so it already embodies typical cumulative exposure), and the ex-smoker residual floor is keyed on total years smoked. A genuine late starter - say, someone who took up smoking at 50 - is scaled down from the lifelong-smoker curve in proportion to their actual dose (years smoked ÷ years they could have smoked since age 20), so they never inherit a lifelong smoker's risk. This replaces an earlier flat amplifier that, combined with the age-varying curve, would have double-counted duration.

Smoking

Smoking follows the same rule as every other factor: the daily ledger shows the per-day cost of harmful behaviors and a clean zero for absent ones. If you've never smoked, no smoking card appears at all. If you've ever smoked, the card asks Did you smoke today? - a Yes logs the full current-smoker daily-rate penalty for that day, a No logs nothing.

The benefit of quitting lives where it really is - in your banked days: the LE difference between your current chronic state and the counterfactual where you'd kept smoking right up to today. That number is shown on your Today page and in your weekly email, and it grows monotonically as your chronic hazard ratio decays from the current-smoker level toward your duration-keyed residual floor along an exponential half-life of 3.5 years (Cho et al, NEJM 2024).

Your chronic state is forgiving of isolated slips: it only moves back toward "current smoker" when you've smoked on 7 or more of the last 14 days. A single Yes after a long clean streak still costs you that day's hours in the ledger, but it doesn't erase your banked days - your recovery curve doesn't reset. The long-term projection age-varies your current-smoker risk by attained age and lands your post-quit floor according to how many years you smoked, as described above.

Activity and Movement

For activity, we model using MET-minutes per week as a single unified metric that combines all aerobic exercise types - cardio (7.0 METs) and walking (3.75 METs). This follows the Lee et al. 2022 dose-response curve, where each workout's contribution is its duration in minutes multiplied by its MET intensity. So, a 30-minute run contributes 210 MET-min; a 30-minute walk contributes 112.5 MET-min. Both count toward the same curve, giving proportional credit regardless of intensity.

We anchor on "cardio" (7.0 METs) and "walking" (3.75 METs) to make the interface simpler. In the future, we may add workout type specific METs if most users use fitness trackers that automatically categorize workouts, or provide KJ/MET-min estimates. But for most people 7.0 is a reasonable estimate, and the precision offered is a false one. Work out in a way that gets your heart moving, for 30-60 minutes a day, and you're maximizing the benefit.

Strength training (7.0 METs) is excluded from the activity HR curve because it has its own separate mortality factor. However, strength MET-minutes do count toward the sedentary PA offset (see below).

Sedentary Behavior: Sit Less, Move More

Prolonged sitting is an independent risk factor for all-cause mortality, separate from exercise. We model this using two complementary findings:

Patterson 2018 (two-slope dose-response): The reference group is people who sit less than 4 hours per day (HR 1.0). Above 4 hours, the risk follows a two-slope pattern: from 4 to 8 hours, each additional hour adds a modest relative risk (HR) of 1.01. Above 8 hours, the slope steepens sharply - each additional hour adds relative risk (HR) 1.04. This means moving from 10 to 12 hours of sitting is much more harmful than moving from 4 to 6 hours.

Ekelund 2016 (physical activity offset): Approximately 60–75 minutes per day of moderate physical activity (~250 MET-min/day) fully eliminates the excess mortality risk from prolonged sitting. At lower Physical Activity levels, the offset is proportional - 125 MET-min/day eliminates about half the sitting penalty. This is why the "Sit Less, Move More" card shows how many more minutes of activity would eliminate your sitting risk for the day.

All workout types contribute to the Physical Activity offset, including strength training. The offset applies to the sedentary penalty only - it doesn't affect the activity factor curve.

Known Limitations & Honest Caveats

This model is a research-grounded estimate, not a medical prediction. We'd rather be honest about where it's approximate than over-claim precision. The most important caveats:

• Period, not cohort, life tables. The baseline survival curves reflect today's death rates, applied unchanged across your whole life. They assume medicine and mortality stay frozen at current levels. Real future cohorts will likely live longer as treatment improves, so the absolute numbers are conservative.
• Factors are treated as independent. We multiply the eight hazard ratios together, which assumes each factor's effect is independent of the others. In reality they interact (e.g. exercise and weight, drinking and sleep). The research on most pairwise interactions is thin, so we use the simpler multiplicative model rather than invent interaction terms.
• One hazard ratio per factor. Each factor collapses to a single HR drawn from population-average studies. Your personal genetics, medical history, medications, and environment can move your real risk substantially in either direction. This is a population model wearing your inputs, not a personalized clinical risk score.
• The disability model is approximate. The healthy-vs-impaired split uses the Sullivan method with disability shift multipliers calibrated on a handful of cohort studies (Klijs 2011, Reuser 2009, Chakravarty 2008). The weight–disability piece has been recalibrated directly from Reuser's sex-split transition data and bounded at the top of their range, but the other factors still use the simpler multiplier and remain directional at the extremes. Read the healthy-years figure as directional, not exact.
• Social connection is a proxy. We approximate the Berkman-Syme social network research with four self-reported inputs (partner, friends, group, activity). Real social integration is richer and harder to measure than four checkboxes. Find people you care about, and spend time with them.
• Recovery curves are modeled, not measured. The "banked days" from quitting smoking, losing weight, or getting active follow exponential decay curves anchored to a few longitudinal studies. The shape is reasonable but your individual recovery trajectory may differ.
• No genetic, environmental, or healthcare factors. The model knows nothing about your family history, pollution exposure, access to care, income, or stress. These are major drivers of real-world life expectancy that simply aren't in this model.

Use this as a directional tool for understanding which habits move the needle most for someone like you - not as a forecast of your actual lifespan.

Research Sources

Every number in this model comes from peer-reviewed epidemiological research. The full citations are organized by topic below.