The Science · Updated

The research behind your number.

Every number hybridAF shows you is built on published research. Some of it is settled science. Some of it is engineering we built on top of that science. Those are different things, and this page keeps them apart, metric by metric, so you can judge each one yourself.

The pattern below repeats. First the validated construct: what decades of physiology actually establish. Then our design choice: the specific formula we wrapped around it. Where the research is strong, we lean on it. Where we made a judgment call, we say so.

Pillar 01

Strength Index

Strong evidence

Your Strength Index starts with the heaviest set you can actually do and estimates the single rep you could grind out from it. We use the Epley formula, 1RM = weight × (1 + reps ÷ 30), across your squat, bench, deadlift and overhead press, for sets of one to twelve reps. Epley began life as a gym wall chart, not a study, but every validation since has found it holds: predicted and measured 1RMs correlate above r = 0.95, and the error stays near ±5% when reps are low.1,2 Past six reps the estimate drifts high, which is why we cap the input.

We then divide that best lift by your bodyweight and grade the ratio against sex-specific standards, from untrained to elite. Relative strength is the honest comparison: a 140 kg deadlift means something different at 60 kg than at 100 kg, and the normative ranges we grade against are drawn from large competition datasets, including one analysis of more than 800,000 entries.3,4,5

Our design choice. Dividing straight by bodyweight (ratio scaling) is simple and slightly favours lighter lifters compared with allometric scaling, which corrects for the fact that strength doesn’t rise linearly with size. We chose ratio scaling for legibility, and we mention the trade-off rather than hide it. The Epley formula is a validated heuristic, not a law.

Pillar 02

Engine Index

Strong evidence

The Engine Index is your VO2max, the peak rate your body can use oxygen, graded against age and sex norms. Of everything we measure, this is the construct with the deepest evidence base. The American Heart Association calls cardiorespiratory fitness a clinical vital sign; higher VO2max tracks strongly with lower mortality, and the relationship is graded, with no plateau where more fitness stops helping.6,7,8 The age and sex cut points we grade against start from the FRIEND fitness registry.9 They match its median for men under 30 and are set stricter than it for women and older age bands, which is a design choice we are reviewing.

We don’t put you in a lab, so your VO2max is a field estimate, not a gas-exchange measurement. Rather than trust one method, we run every signal you give us through the published equation built for it:

  • Your fastest runs. Daniels and Gilbert’s VDOT reads a VO2max off pace and duration.10 Hills are converted to flat-equivalent distance using the measured cost of running on slopes.11 Heart rate decides whether the effort was all-out. If it wasn’t, the run only proves you are at least that fit.
  • Your steady runs with heart rate. The share of your heart-rate reserve you use tracks the share of your oxygen reserve,12,13 so a steady pace and the heart rate it costs you extrapolate to a ceiling. The oxygen cost of the pace comes from Daniels’ running curve, not the common ACSM equation, which overestimates trained athletes.14 Your max heart rate is the highest you’ve recorded; if you haven’t gone near it, we fall back to an age prediction,15 which carries a typical error of about 11 beats a minute,16 and the range widens to match.
  • Your hard rides with power. Best five-minute power per kilo of bodyweight, counted only when your heart rate shows the effort was near-maximal.17
  • Your watch’s own VO2max. One input, not the answer. Exercise-based watch estimates land within roughly 10 ml/kg/min of lab values for 95% of people,18 and one Apple Watch study found readings about 6 below the lab on average,19 so the watch counts for less than a well-measured run.
  • Your resting heart rate. The ratio of max to resting heart rate predicts VO2max in trained men and women in a lab.20,21 We compute it, but we don’t count it: on our own athletes it read higher than their training data every time we could check, most likely because a wrist “resting heart rate” runs below the rested, lying-down value the method was built on.

Our design choice. The way we combine them is ours. Each method carries an error taken from its validation research, except the run-based VDOT error, which no study we could verify reports against lab VO2max, so we set it ourselves. Estimates are weighted by the inverse of their error, older data counts for less, and the range we show assumes the methods partly share their mistakes, since they all read the same heart. When a heart-rate confirmed hard effort disagrees sharply with everything else, the hard effort wins: what you demonstrably ran outranks what a heart rate implies. Every estimate is kept with its range and the methods behind it, and the app names those methods next to your Engine Index. VO2max here is a training metric, not a medical test.

The same equations run in the browser on the VO2max calculator: type in a race, a steady run or a ride and it shows the estimate, its range and which methods counted.

Metric 03

Fitness Age

Moderate evidence

Fitness Age answers a simpler question than VO2max: your aerobic fitness is typical of someone how old? It’s the same measurement translated into a number people feel. The idea comes from the HUNT Fitness Study in Norway, which mapped VO2max onto age across tens of thousands of people, and it’s the basis for NTNU’s widely used fitness calculator.22,23 We grade it using the same age and sex mortality data behind the Engine Index.7

Our design choice. Fitness Age is a communication device, not a new measurement. The original HUNT model was non-exercise (built from questionnaire data); ours adapts the same idea from your running. Treat it as a motivating translation of your VO2max, not a second, independent reading.

The Flagship 04

Hybrid Score

Our design choice

The Hybrid Score fuses the two pillars into one number. Both are graded 0–100 on the same percentile ladder, and we combine them with a harmonic mean rather than an average. That choice matters: a harmonic mean is dragged toward the smaller of the two. Be a strong lifter with no engine, or a runner who can’t squat, and the score punishes the gap. You only score well by being good at both.

The reasoning is grounded. Muscular strength independently predicts mortality, and so does cardiorespiratory fitness, and the people who are high in both carry the lowest risk of all.24,25,26,27 A score that rewards balance rather than a single spike reflects that finding. The ingredients are evidence-based.

Our design choice, stated plainly. No published instrument validates a harmonic mean of strength × engine as a health or performance metric. We built it because it captures something real, that lopsided athletes are worse off than balanced ones, but the composite is our engineering, not a clinical tool. The pillars are validated; the way we blend them is a considered opinion.

Daily signal 05

Recovery

Strong evidence

Your Recovery score (0–100) reads how ready your nervous system is, mostly from heart rate variability. The research mostly uses rMSSD, the beat-to-beat measure that best reflects parasympathetic (rest-and-recover) activity. Apple Health records a related measure, SDNN, and that is what we read from an Apple Watch. The signal that matters isn’t your HRV against the population; it’s your HRV against your own baseline. A value that’s low for you is the alarm. Comparing you with your own recent normal is the method the research endorses.28,29 Our baseline today is the average of every reading we hold for you, not a rolling window, which is a design choice we are reviewing. Resting heart rate adds a weaker secondary read, and sleep rounds it out.30,31

Our design choice. We weight the three inputs 45% HRV, 30% resting heart rate, 25% sleep. Those weights are a heuristic: a sensible ordering of signal strength, not a blend anyone has validated. The parts are strongly evidenced; the exact recipe is our prior, and we tune it as we learn from real athlete data.

Daily signal 06

Strain

Moderate evidence

Strain compresses a day of load into a single number from 0 to 100. It is built from the day’s active energy, the calories your iPhone and Apple Watch record in Apple Health. The principle underneath it is old and solid: load is intensity multiplied by duration, and the cost of effort rises non-linearly: the last hard interval taxes you far more than the first.32,33,34 That’s why the scale is logarithmic rather than linear: it stretches out the top end where the real damage accumulates.

Our design choice. The curve and the cut points (Rest, Easy, Steady, Hard, Max) are our calibration, not a published standard. And the input has a known blind spot: active energy is itself estimated by the watch from heart rate and movement, so it under-weights short, sharp, intermittent work (think heavy lifting or sprint intervals) where the cost outruns what heart rate reports. Read Strain as a strong directional signal, not a precise fuel gauge.

How it acts 07

Readiness & adaptive rescheduling

Strong evidence

Readiness turns your recovery into a decision: green, yellow or red. When it’s low, the app moves your hard sessions off that day and protects it for easier work. This is where the numbers do something. Training that adapts to daily readiness, HRV-guided rather than fixed on a calendar, performs at least as well as a rigid plan, and it produces fewer non-responders and fewer athletes who stall or regress.35,36,37,38

The honest read. The gains here are about individualisation, not raw ceiling. Several studies find the effect on peak aerobic capacity is small or not statistically significant. The win is that more people improve and fewer break down, not that the best-case number goes higher.39 Most of this research is in endurance athletes, so we apply it to strength work with appropriate caution.

Underneath it all 08

How we grade, and how the coach thinks

Our design choice

Almost every number here is a percentile. A raw lift or a raw VO2max means little on its own, so we grade it against normative data for your age and sex and tell you where you land: untrained to elite, or 0 to 100. That’s what makes two very different athletes comparable on one screen.

Our design choice. Work capacity is graded on how often you train and how much you move, and loaded circuit work counts toward the second half: a sled push or a sandbag carry is real volume that no set-and-rep log records. Travel under load carries no rep count, so we convert it at five metres per rep before adding it to barbell tonnage. That rate is our prior, not a measured equivalence, and we set it conservatively: a full race simulation lands near six tonnes, a hard day rather than a shortcut past athletes who lift.

Our design choice, and the weakest claim on this page. The what-if projection on your score page is a model, not a forecast, and nothing validates it. It reads the best sustained rate your own score has actually climbed over any stretch of three weeks or more in your history, divides that by how often you were training each pillar at the time, and carries the result forward eight weeks. Gains then slow as you climb, because the grading ladder itself flattens near the top: the same physical improvement is worth fewer index points there, and the projection spends its gain band by band at each band’s real price. With under three weeks of history, or a pillar that has never climbed, it shows you nothing rather than a guess. Treat the number as an estimate of where the model lands, never as a promise about your body.

The AI coach reasons from the same evidence base rather than improvising. It reads your scores, your recovery and your calendar, and reasons over a curated knowledge base of the research summarised on this page, so its advice stays anchored to what’s measured and what’s published.

The bottom line

What’s proven, and what we designed.

Two things on this page are validated science in their own right: VO2max predicts long-term health, and training that adapts to HRV-guided readiness beats a rigid plan for individualising load. We didn’t invent those; we built on them.

Four things are ours. The harmonic-mean Hybrid Score, the way we weigh and combine the VO2max methods, the 45/30/25 recovery weights, and the strain curve are engineering decisions the science informs but doesn’t certify as-built. We think they’re good decisions, backed by the evidence for their ingredients, and we’d rather tell you exactly where the line sits than blur it. That’s the whole point of this page.

References

  1. 01LeSuer DA, McCormick JH, Mayhew JL, et al. The accuracy of prediction equations for estimating 1-RM performance in the bench press, squat, and deadlift. J Strength Cond Res. 1997;11(4):211–213. View source
  2. 02Macht JW, Abel MG, Mullineaux DR, Yates JW. Development of 1RM prediction equations for bench press in moderately trained men. J Strength Cond Res. 2016;30(10):2901–2906. View source
  3. 03van den Hoek DJ, Beaumont PL, van den Hoek AK, Owen PJ, Garrett JM, Buhmann R, Latella C. Normative data for the squat, bench press and deadlift exercises in powerlifting: data from 809,986 competition entries. J Sci Med Sport. 2024;27(10):734–742. View source
  4. 04Bartolomei S, Grillone G, Di Michele R, Cortesi M. A comparison between male and female athletes in relative strength and power performances. J Funct Morphol Kinesiol. 2021;6(1):17. View source
  5. 05Falch HN, Haugen ME, Larsen S, van den Tillaar R. Association of strength performance in bench press and squat with anthropometric variables between resistance-trained males and females. J Funct Morphol Kinesiol. 2023;8(1):19. View source
  6. 06Ross R, Blair SN, Arena R, et al. Importance of assessing cardiorespiratory fitness in clinical practice: a case for fitness as a clinical vital sign. A scientific statement from the American Heart Association. Circulation. 2016;134(24):e653–e699. View source
  7. 07Mandsager K, Harb S, Cremer P, et al. Association of cardiorespiratory fitness with long-term mortality among adults undergoing exercise treadmill testing. JAMA Netw Open. 2018;1(6):e183605. View source
  8. 08Han M, Qie R, Shi X, et al. Cardiorespiratory fitness and mortality from all causes, cardiovascular disease and cancer: dose-response meta-analysis of cohort studies. Br J Sports Med. 2022;56(13):733–739. View source
  9. 09Kaminsky LA, Arena R, Myers J. Reference standards for cardiorespiratory fitness measured with cardiopulmonary exercise testing: data from the Fitness Registry and the Importance of Exercise National Database. Mayo Clin Proc. 2015;90(11):1515–1523. View source
  10. 10Daniels J, Gilbert J. Oxygen Power: Performance Tables for Distance Runners. Tempe, AZ: self-published; 1979.
  11. 11Minetti AE, Moia C, Roi GS, Susta D, Ferretti G. Energy cost of walking and running at extreme uphill and downhill slopes. J Appl Physiol. 2002;93(3):1039–1046. View source
  12. 12Swain DP, Leutholtz BC. Heart rate reserve is equivalent to %VO2 reserve, not to %VO2max. Med Sci Sports Exerc. 1997;29(3):410–414. View source
  13. 13Swain DP, Leutholtz BC, King ME, Haas LA, Branch JD. Relationship between % heart rate reserve and % VO2 reserve in treadmill exercise. Med Sci Sports Exerc. 1998;30(2):318–321. View source
  14. 14Koutlianos N, Dimitros E, Metaxas T, Cansiz M, Deligiannis A, Kouidi E. Indirect estimation of VO2max in athletes by ACSM's equation: valid or not? Hippokratia. 2013;17(2):136–140. View source
  15. 15Tanaka H, Monahan KD, Seals DR. Age-predicted maximal heart rate revisited. J Am Coll Cardiol. 2001;37(1):153–156. View source
  16. 16Nes BM, Janszky I, Wisløff U, Støylen A, Karlsen T. Age-predicted maximal heart rate in healthy subjects: the HUNT Fitness Study. Scand J Med Sci Sports. 2013;23(6):697–704. View source
  17. 17Sitko S, Cirer-Sastre R, Corbi F, López-Laval I. Five-minute power-based test to predict maximal oxygen consumption in road cycling. Int J Sports Physiol Perform. 2022;17(1):9–15. View source
  18. 18Molina-Garcia P, Notbohm HL, Schumann M, et al. Validity of estimating the maximal oxygen consumption by consumer wearables: a systematic review with meta-analysis and expert statement of the INTERLIVE network. Sports Med. 2022;52(7):1577–1597. View source
  19. 19Lambe R, O'Grady B, Baldwin M, Doherty C. Investigating the accuracy of Apple Watch VO2 max measurements: a validation study. PLoS One. 2025;20(5):e0323741. View source
  20. 20Uth N, Sørensen H, Overgaard K, Pedersen PK. Estimation of VO2max from the ratio between HRmax and HRrest: the Heart Rate Ratio Method. Eur J Appl Physiol. 2004;91(1):111–115. View source
  21. 21Uth N. Gender difference in the proportionality factor between the mass specific VO2max and the ratio between HRmax and HRrest. Int J Sports Med. 2005;26(9):763–767. View source
  22. 22Nes BM, Janszky I, Vatten LJ, Nilsen TI, Aspenes ST, Wisløff U. Estimating VO2peak from a nonexercise prediction model: the HUNT Study, Norway. Med Sci Sports Exerc. 2011;43(11):2024–2030. View source
  23. 23NTNU Cardiac Exercise Research Group (CERG). World Fitness Level / VO2max calculator. View source
  24. 24Volaklis KA, Halle M, Meisinger C. Muscular strength as a strong predictor of mortality: a narrative review. Eur J Intern Med. 2015;26(5):303–310. View source
  25. 25Kim Y, White T, Wijndaele K, et al. The combination of cardiorespiratory fitness and muscle strength, and mortality risk. Eur J Epidemiol. 2018;33(10):953–964. View source
  26. 26Leong DP, Teo KK, Rangarajan S, et al. Prognostic value of grip strength: findings from the Prospective Urban Rural Epidemiology (PURE) study. Lancet. 2015;386(9990):266–273. View source
  27. 27Schnohr P, O'Keefe JH, Lavie CJ, Suetta C, Jensen GB, Marott JL. Comparison of muscle strength and cardiorespiratory fitness in relation to cardiovascular and all-cause mortality: the Copenhagen City Heart Study. Mayo Clin Proc. 2025;100(3):488–498. View source
  28. 28Plews DJ, Laursen PB, Stanley J, Kilding AE, Buchheit M. Training adaptation and heart rate variability in elite endurance athletes: opening the door to effective monitoring. Sports Med. 2013;43(9):773–781. View source
  29. 29Plews DJ, Laursen PB, Kilding AE, Buchheit M. Heart rate variability in elite triathletes, is variation in variability the key to effective training? A case comparison. Eur J Appl Physiol. 2012;112(11):3729–3741. View source
  30. 30Bosquet L, Merkari S, Arvisais D, Aubert AE. Is heart rate a convenient tool to monitor over-reaching? A systematic review of the literature. Br J Sports Med. 2008;42(9):709–714. View source
  31. 31Manresa-Rocamora A, Flatt AA, Casanova-Lizón A, et al. Heart rate-based indices to detect parasympathetic hyperactivity in functionally overreached athletes. A meta-analysis. Scand J Med Sci Sports. 2021;31(6):1164–1182. View source
  32. 32Haddad M, Stylianides G, Djaoui L, Dellal A, Chamari K. Session-RPE method for training load monitoring: validity, ecological usefulness, and influencing factors. Front Neurosci. 2017;11:612. View source
  33. 33Foster C, Florhaug JA, Franklin J, et al. A new approach to monitoring exercise training. J Strength Cond Res. 2001;15(1):109–115. View source
  34. 34Borg E, Kaijser L. A comparison between three rating scales for perceived exertion and two different work tests. Scand J Med Sci Sports. 2006;16(1):57–69. View source
  35. 35Vesterinen V, Nummela A, Heikura I, et al. Individual endurance training prescription with heart rate variability. Med Sci Sports Exerc. 2016;48(7):1347–1354. View source
  36. 36Javaloyes A, Sarabia JM, Lamberts RP, Moya-Ramon M. Training prescription guided by heart rate variability in cycling. Int J Sports Physiol Perform. 2019;14(1):23–32. View source
  37. 37Javaloyes A, Sarabia JM, Lamberts RP, Plews D, Moya-Ramon M. Training prescription guided by heart rate variability vs. block periodization in well-trained cyclists. J Strength Cond Res. 2020;34(6):1511–1518. View source
  38. 38Manresa-Rocamora A, Sarabia JM, Javaloyes A, Flatt AA, Moya-Ramón M. Heart rate variability-guided training for enhancing cardiac-vagal modulation, aerobic fitness, and endurance performance: a methodological systematic review with meta-analysis. Int J Environ Res Public Health. 2021;18(19):10299. View source
  39. 39Granero-Gallegos A, González-Quílez A, Plews D, Carrasco-Poyatos M. HRV-based training for improving VO2max in endurance athletes. A systematic review with meta-analysis. Int J Environ Res Public Health. 2020;17(21):7999. View source