Lifted

The science behind Lifted

What the app does, why, and how sure we are.

This is the single source of truth for every rule in Lifted. The same file is served to your AI coach through the connector (lifted://evidence), published at liftedcoach.app/science, and linked from inside the app wherever a rule is explained. If the app and this document ever disagree, that is a bug and the document wins.

Every citation below was checked against the paper before publication — the DOI resolves and the abstract supports the sentence it is attached to. Where we could not verify a claim, it was dropped or downgraded rather than left in. Full references with DOIs are at the end.

Confidence vocabulary, used throughout:

• strong — multiple randomised trials or a meta-analysis point the same way
• moderate — consistent trials or cohort studies, or a meta-analysis with a small effect
• weak — expert practice, mechanistic reasoning, or contested findings
• judgement — a product decision we made on top of the evidence, and say so

Last revised 2026-09-08. Changes are listed at the end.

1. Measuring recovery

1.1 HRV as a recovery signal — moderate

Resting vagally-mediated heart rate variability (RMSSD) falls with accumulated fatigue, illness, poor sleep and alcohol, and recovers as the athlete does. The relationship holds at the group level; day-to-day noise in one person is large, which is why nothing in Lifted reads a single morning on its own.

What the app does: scores each morning against your own history, never against a population norm.

— Plews et al. 2013 [1]; Buchheit 2014 [2].

1.2 Log-transform and a rolling baseline — strong for the method

RMSSD is skewed; its natural log is approximately normal. A 7-day rolling mean of lnRMSSD is the standard way to see the trend through the noise, and the case that made it standard is instructive: in two elite triathletes tracked daily for 77 days, the 7-day rolling lnRMSSD declined toward a key race in the athlete who was later diagnosed as non-functionally over-reached, and stayed flat in the one who was not [3].

What the app does: compares this morning to your rolling 7-reading baseline, with normal day-to-day spread estimated from your last 60.

— Plews et al. 2012 [3]; Plews et al. 2013 [1].

1.3 The smallest worthwhile change — moderate

A change smaller than about half your own standard deviation is noise. The "normal" band in Lifted is your baseline ± 0.5 SD of lnRMSSD; suppressed starts below it. The convention comes from performance-testing statistics, where 0.2 SD is the smallest worthwhile change for a group and 0.5 SD is a conservative threshold for an individual [4].

— Hopkins 2004 [4]; Buchheit 2014 [2].

1.4 Resting heart rate — moderate

An elevated morning resting heart rate tracks incomplete recovery and illness, and is less noisy than HRV. It is weighted second in the score, and it is the tie-breaker in §1.8.

When there is no HRV at all — a Whoop, say, which writes resting HR and sleep to Apple Health but not HRV — resting HR becomes the band: suppressed at one standard deviation above the person's own baseline, or half a standard deviation when they also report feeling a full SD worse than usual. Never "elevated": a low resting HR is not evidence of super-compensation. The +1 SD bar is ours; Buchheit's review supports RHR as a real but blunter marker than HRV, which is why the bar is higher than the 0.5 SD used for HRV.

— Buchheit 2014 [2].

1.5 How you feel — moderate for sensitivity, judgement for weighting

A systematic review of athlete monitoring found that simple self-reported wellness measures responded to training load more consistently than the commonly used objective measures, and often earlier [5]. Lifted asks one question — 1 to 10 — and lets it move the score.

What the app does, and deliberately does not: whenever the morning has an objective term — HRV or resting HR — feel moves the score, never the band. The band is what training rules key on, and a bad mood is not evidence of a suppressed nervous system. That split is a judgement, not a finding.

With nothing but feel (no strap, no wearable), feel is the whole reading and therefore the band too. The bar is deliberately strict and doubly gated: a 3/10 or lower, or a 4 two mornings running, and at least one standard deviation below the person's own baseline. Both conditions because a single number from a groggy person is noisier than an ECG, and because a person whose normal is 4 must not trip the rule every other day. The review above is the reason a feel-only tier is offered at all rather than a "no data" screen: self-report tracked training load as well as or better than the objective markers in the studies it pooled. The thresholds are ours.

— Saw, Main & Gastin 2016 [5].

1.6 How long before a score means anything — judgement

No score before 5 mornings. "Calibrating" until 14, and a calibrating score never drives an adjustment. These numbers sit on top of the smallest- worthwhile-change literature but are ours: with five days the spread estimate is unreliable, so it is floored at a physiological minimum rather than letting a 5-day SD of 0.02 turn ordinary noise into a "suppressed" verdict.

1.7 Measurement: what a valid HRV reading needs — strong

RMSSD needs true beat-to-beat (RR) intervals. A chest strap that transmits RR is validated against ECG: the Polar H10 held 99.4% RR signal quality against a Holter ECG through high-intensity activity [6]. Optical wrist and finger sensors are not validated for resting HRV to that standard, and cheaper straps add timing jitter that inflates the number.

Wearables that write overnight HRV to Apple Health report SDNN, not RMSSD. The two are on different scales and Lifted never mixes them: a Health reading is scored only against other Health readings, a strap reading only against strap readings.

— Gilgen-Ammann, Schweizer & Wyss 2019 [6].

1.8 HRV alone is ambiguous; resting HR disambiguates — moderate

Suppressed HRV with a normal resting HR is usually training fatigue. Suppressed HRV together with an elevated resting HR skews toward illness or genuine overreaching. And in well-trained endurance athletes HRV can rise, not fall, under heavy load — parasympathetic saturation — which is one more reason never to read HRV without the resting heart rate beside it [1, 2].

What the app does: the illness state (§6) only fires on the pair.

2. Letting recovery change training

2.1 HRV-guided training works, in one direction — moderate

Three trials prescribed training day by day from morning HRV and compared it with a fixed plan: moderately fit men [7], recreational runners [8], and well-trained cyclists [9]. In each, the HRV-guided group matched or beat the fixed plan on performance. A 2021 meta-analysis of eight such studies (198 participants) found a medium positive effect on submaximal physiological measures, g = 0.30 [10].

The part that shapes Lifted: in these protocols the guided groups did fewer hard sessions, not more. The benefit came from taking hard days away on suppressed mornings. None of the trials added intensity because HRV was high.

What the app does: every recovery rule is validated as reduce-only. There is no rule, and no way to write one, that adds load because a number looked good.

— Kiviniemi et al. 2007 [7]; Vesterinen et al. 2016 [8]; Javaloyes et al. 2019 [9]; Düking et al. 2021 [10].

2.2 The honest gap: that evidence is endurance, not lifting

There is very little HRV-guided resistance training research. Strength appears fairly robust to a single bad morning. So for a lifter, Lifted does not read a suppressed morning as "skip it"; it reads it as "cut the part of the session that generates fatigue without driving the adaptation" — the volume and the accessories — and keep the top set. The two sections below are the reason that is the right cut.

2.3 Stopping a set early is a real intervention — moderate

Two randomised training studies compared programmes that were identical except for how far into fatigue each set went, measured as loss of bar velocity. Stopping at 20% velocity loss produced strength and athletic performance gains equal to or better than continuing to 40%, with much less fatigue and less fibre-type shift toward slower fibres [11]. A follow-up with four thresholds (0, 10, 20, 40%) found the same pattern across strength, jump and sprint outcomes [12].

Practically: the reps at the end of a set, where the bar has visibly slowed, cost the most and give the least.

What the app does: "Run it with a brake" — same loads, stop each set when reps slow or at about two reps in reserve. It has no progression cost.

— Pareja-Blanco et al. 2017 [11]; Pareja-Blanco et al. 2020 [12].

2.4 Reps in reserve as the in-session gauge — moderate

Repetitions-in-reserve ratings track bar velocity and percentage of 1RM closely in trained lifters (r ≈ 0.88–0.91 between %1RM and RPE across the squat, bench and deadlift in powerlifters) [14], and the RIR-based RPE scale is reliable enough to detect a grind [13].

What the app does: an optional RIR entry per set; a lift configured with an RIR brake holds instead of progressing after a hit at or below the brake. Progressing after a hit at RIR 0 usually produces a miss next time.

— Zourdos et al. 2016 [13]; Helms et al. 2017 [14].

2.5 A reduced day holds the plan — judgement, and a correction

Progression in Lifted reads the log. Before September 2026 that meant a day reduced by a recovery rule to 90% and hit prescribed 0.9 × w + increment next time — the lifter was stepped down for doing what the app told them to do. That was a bug. A day deliberately made easier is not evidence about the planned load in either direction, so it now holds the plan: no progression, no miss, no deload count. A genuine stall still shows on the next full day.

2.6 A skipped session is skipped, not owed — strong

Three findings point the same way.

• When weekly volume is held equal, how many sessions it is spread across makes no significant difference to hypertrophy (25 studies) [15] or to strength (volume-equated subgroup, p = 0.42) [16]. A session is not a unit that needs "making up"; the week is.
• One missed session is far below the detraining floor. Short-term detraining reviews put measurable losses at weeks of insufficient stimulus, not days [17], and strength and muscle can be maintained for months on a fraction of the training that built them [18].
• Stacking two heavy sessions for the same muscles inside 48 hours, on a week you skipped because you were not recovered, is adding load in a fatigued state — the exact thing §2.1 exists to prevent.

What the app does: a skipped session comes round again next week at the same prescription. The app never proposes a make-up session.

— Schoenfeld, Grgic & Krieger 2019 [15]; Grgic et al. 2018 [16]; Mujika & Padilla 2000 [17]; Bickel, Cross & Bamman 2011 [18].

2.7 What does cost progress: weekly volume drifting — moderate

Weekly hard sets show a dose-response with muscle growth — each additional weekly set associated with a further 0.37% gain across 15 studies [19]. So it is a pattern of skipped sessions, not any one of them, that shows up in results.

What the app does: tracks hard sets per group per week against your own trailing average and says something once, only when the trend breaks. It does not tell you to train more; it tells you the trend has moved.

— Schoenfeld, Ogborn & Krieger 2017 [19].

3. Strength progression

3.1 Progressive overload — strong

Load, reps or sets must rise over time for continued gains; the specific scheme matters less than the consistency [20].

3.2 Linear progression — practice

Hit → add the increment; miss → repeat; two misses → deload 10%. The standard novice–intermediate scheme; the deload after repeated misses is the usual answer to a stall [21].

3.3 Double progression — practice

Reps climb within a range on hits; weight moves only when the top of the range is hit. Used where the smallest available jump is large relative to the load.

3.4 Volume day at 80% of heavy, plus two reps — practice

A lighter repeat gives frequency without a second heavy exposure.

— Schoenfeld et al. 2017 [20]; Rippetoe & Baker [21].

4. Endurance

4.1 Zones — practice

%HRmax bands for load; heart-rate reserve or threshold-based targets when you prefer them. None is "correct"; consistency matters more than the method.

— Karvonen, Kentala & Mustala 1957 [22]; Friel [23].

4.2 Training load (TRIMP) — moderate

Banister's TRIMP and Edwards' zone score are the established heart-rate-based internal load measures. Both are computed; neither is a target.

— Banister 1991 [24]; Edwards 1993 [25].

4.3 Acute:chronic workload ratio — weak, contested

Shown as a trend only; the plan does not react to it. The original claim that it predicts injury has not held up: no study has properly estimated a causal effect, and manipulating the ratio to change injury rates is an over-interpretation of the data [27].

— Gabbett 2016 [26]; Impellizzeri et al. 2020 [27].

4.4 Aerobic decoupling — weak

The "under 5% drift on a long steady effort" heuristic is popular and plausible; the mechanism of cardiovascular drift is well described [28], the training heuristic is practice [23]. Shown per session, never used to prescribe.

5. What actually helps recovery

Ranked by effect size and how sure we are. The app offers at most three of these on a low morning, and it never claims any of them will move your score.

5.1 The one we tell you not to do — moderate

Cold water immersion after lifting. Twelve weeks of strength training with 10 minutes of cold water after each session produced smaller gains in strength and muscle mass than active recovery, and blunted the acute anabolic signalling that drives adaptation [35]. A second trial found the hypertrophy effect again, though strength was not impaired in that one [36]. It is the most popular recovery ritual in the culture and it works against a strength goal.

The evidence for hypertrophy is consistent; for strength it is mixed. Either way, if you lift for size or strength, keep the ice bath for a rest day.

— Roberts et al. 2015 [35]; Fyfe et al. 2019 [36].

6. Illness — safety, not coaching

When suppressed HRV meets an elevated resting heart rate, the app stops offering a menu and states a position: don't train hard today, and see a doctor if you have a fever.

The basis is precautionary rather than a trial: the European Society of Cardiology's sports guidance recommends abstaining from sport while myocardial inflammation is active, and notes that in most myocarditis cases the presentation is preceded by a prodrome — fever in about 60%, respiratory or gastrointestinal symptoms in others [37]. Nobody can tell from a heart rate strap whether a given febrile morning is one of those. The cost of resting a day you did not need to rest is one session; the cost of the other error is not. — practice / precautionary

— Pelliccia et al. 2020 [37].

7. What Lifted deliberately does not do

• No population norms. "Your RMSSD of 45 is good" is not a thing we will ever say; healthy adults span roughly 15–120 ms.
• No readiness-based increases. Nothing adds load because a score is high.
• No injury prediction. ACWR is a trend, not a forecast (§4.3).
• No model-computed numbers. Your AI may propose a program; the engine validates it and you accept it. The AI never writes a weight.
• No claim that any recovery action raises your score. The app reports what happened after you did it (§8); it does not promise.
• No diagnosis. The illness state points at a doctor and stops.

8. How the app learns what works for you

Once you log recovery actions (§5), Lifted compares the next morning's recovery on nights you did the thing with nights you did not. This is observational data on one person, and the app treats it that way:

• nothing is shown until there are at least 15 observations on each side;
• the result is a difference with a spread, never a p-value, and never the word "causes";
• one line under the table says exactly this.

It will not tell you what works for people. It may, after a few months, tell you what has tended to precede your better mornings.

Changelog

• 2026-09-08 — recovery tiers. §1.4 and §1.5 gain the rhr-only and feel-only tiers: a morning is scored from whatever it has, with tier-specific bands and the bars stated as judgement.

• 2026-09-08 — Rewritten as the single evidence document. Added §2.2–2.7 (the lifting gap, velocity loss, RIR, the reduced-day correction, skipped sessions, weekly volume), §5 (recovery levers, cold water), §6 (illness), §8. All citations re-verified against DOIs.

• 2026-09-07 — First version, covering the recovery score, reduce-only rules, progression, TRIMP and ACWR.

References

1. Plews 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. https://doi.org/10.1007/s40279-013-0071-8
2. Buchheit M. Monitoring training status with HR measures: do all roads lead to Rome? Front Physiol. 2014;5:73. https://doi.org/10.3389/fphys.2014.00073
3. Plews 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. https://doi.org/10.1007/s00421-012-2354-4
4. Hopkins WG. How to interpret changes in an athletic performance test. Sportscience. 2004;8:1–7. https://www.sportsci.org/jour/04/wghtests.htm
5. Saw AE, Main LC, Gastin PB. Monitoring the athlete training response: subjective self-reported measures trump commonly used objective measures: a systematic review. Br J Sports Med. 2016;50(5):281–291. https://doi.org/10.1136/bjsports-2015-094758
6. Gilgen-Ammann R, Schweizer T, Wyss T. RR interval signal quality of a heart rate monitor and an ECG Holter at rest and during exercise. Eur J Appl Physiol. 2019;119(7):1525–1532. https://doi.org/10.1007/s00421-019-04142-5
7. Kiviniemi AM, Hautala AJ, Kinnunen H, Tulppo MP. Endurance training guided individually by daily heart rate variability measurements. Eur J Appl Physiol. 2007;101(6):743–751. https://doi.org/10.1007/s00421-007-0552-2
8. Vesterinen V, Nummela A, Heikura I, et al. Individual endurance training prescription with heart rate variability. Med Sci Sports Exerc. 2016;48(7):1347–1354. https://doi.org/10.1249/MSS.0000000000000910
9. Javaloyes 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. https://doi.org/10.1123/ijspp.2018-0122
10. Düking P, Zinner C, Trabelsi K, et al. Monitoring and adapting endurance training on the basis of heart rate variability monitored by wearable technologies: a systematic review with meta-analysis. J Sci Med Sport. 2021;24(11):1180–1192. https://doi.org/10.1016/j.jsams.2021.04.012
11. Pareja-Blanco F, Rodríguez-Rosell D, Sánchez-Medina L, et al. Effects of velocity loss during resistance training on athletic performance, strength gains and muscle adaptations. Scand J Med Sci Sports. 2017;27(7):724–735. https://doi.org/10.1111/sms.12678
12. Pareja-Blanco F, Alcazar J, Sánchez-Valdepeñas J, et al. Velocity loss as a critical variable determining the adaptations to strength training. Med Sci Sports Exerc. 2020;52(8):1752–1762. https://doi.org/10.1249/MSS.0000000000002295
13. Zourdos MC, Klemp A, Dolan C, et al. Novel resistance training–specific rating of perceived exertion scale measuring repetitions in reserve. J Strength Cond Res. 2016;30(1):267–275. https://doi.org/10.1519/JSC.0000000000001049
14. Helms ER, Storey A, Cross MR, et al. RPE and velocity relationships for the back squat, bench press, and deadlift in powerlifters. J Strength Cond Res. 2017;31(2):292–297. https://doi.org/10.1519/JSC.0000000000001517
15. Schoenfeld BJ, Grgic J, Krieger J. How many times per week should a muscle be trained to maximize muscle hypertrophy? A systematic review and meta-analysis of studies examining the effects of resistance training frequency. J Sports Sci. 2019;37(11):1286–1295. https://doi.org/10.1080/02640414.2018.1555906
16. Grgic J, Schoenfeld BJ, Davies TB, Lazinica B, Krieger JW, Pedisic Z. Effect of resistance training frequency on gains in muscular strength: a systematic review and meta-analysis. Sports Med. 2018;48(5):1207–1220. https://doi.org/10.1007/s40279-018-0872-x
17. Mujika I, Padilla S. Detraining: loss of training-induced physiological and performance adaptations. Part I: short term insufficient training stimulus. Sports Med. 2000;30(2):79–87. https://doi.org/10.2165/00007256-200030020-00002
18. Bickel CS, Cross JM, Bamman MM. Exercise dosing to retain resistance training adaptations in young and older adults. Med Sci Sports Exerc. 2011;43(7):1177–1187. https://doi.org/10.1249/MSS.0b013e318207c15d
19. Schoenfeld BJ, Ogborn D, Krieger JW. Dose-response relationship between weekly resistance training volume and increases in muscle mass: a systematic review and meta-analysis. J Sports Sci. 2017;35(11):1073–1082. https://doi.org/10.1080/02640414.2016.1210197
20. Schoenfeld BJ, Grgic J, Ogborn D, Krieger JW. Strength and hypertrophy adaptations between low- vs. high-load resistance training: a systematic review and meta-analysis. J Strength Cond Res. 2017;31(12):3508–3523. https://doi.org/10.1519/JSC.0000000000002200
21. Rippetoe M, Baker A. Practical Programming for Strength Training. 3rd ed. Aasgaard; 2013.
22. Karvonen MJ, Kentala E, Mustala O. The effects of training on heart rate: a longitudinal study. Ann Med Exp Biol Fenn. 1957;35(3):307–315.
23. Friel J. The Triathlete's Training Bible. 4th ed. VeloPress; 2016.
24. Banister EW. Modeling elite athletic performance. In: MacDougall JD, Wenger HA, Green HJ, eds. Physiological Testing of the High-Performance Athlete. 2nd ed. Human Kinetics; 1991:403–424.
25. Edwards S. The Heart Rate Monitor Book. Polar Electro Oy; 1993.
26. Gabbett TJ. The training–injury prevention paradox: should athletes be training smarter and harder? Br J Sports Med. 2016;50(5):273–280. https://doi.org/10.1136/bjsports-2015-095788
27. Impellizzeri FM, Tenan MS, Kempton T, Novak A, Coutts AJ. Acute:chronic workload ratio: conceptual issues and fundamental pitfalls. Int J Sports Physiol Perform. 2020;15(6):907–913. https://doi.org/10.1123/ijspp.2019-0864
28. Coyle EF, González-Alonso J. Cardiovascular drift during prolonged exercise: new perspectives. Exerc Sport Sci Rev. 2001;29(2):88–92. https://doi.org/10.1097/00003677-200104000-00009
29. Pietilä J, Helander E, Korhonen I, Myllymäki T, Kujala UM, Lindholm H. Acute effect of alcohol intake on cardiovascular autonomic regulation during the first hours of sleep in a large real-world sample of Finnish employees: observational study. JMIR Ment Health. 2018;5(1):e23. https://doi.org/10.2196/mental.9519
30. Zhang S, Niu X, Ma J, Wei X, Zhang J, Du W. Effects of sleep deprivation on heart rate variability: a systematic review and meta-analysis. Front Neurol. 2025;16:1556784. https://doi.org/10.3389/fneur.2025.1556784
31. Laborde S, Allen MS, Borges U, et al. Effects of voluntary slow breathing on heart rate and heart rate variability: a systematic review and a meta-analysis. Neurosci Biobehav Rev. 2022;138:104711. https://doi.org/10.1016/j.neubiorev.2022.104711
32. Lehrer PM, Gevirtz R. Heart rate variability biofeedback: how and why does it work? Front Psychol. 2014;5:756. https://doi.org/10.3389/fpsyg.2014.00756
33. Stults-Kolehmainen MA, Bartholomew JB, Sinha R. Chronic psychological stress impairs recovery of muscular function and somatic sensations over a 96-hour period. J Strength Cond Res. 2014;28(7):2007–2017. https://doi.org/10.1519/JSC.0000000000000335
34. Mountjoy M, Ackerman KE, Bailey DM, et al. 2023 International Olympic Committee's (IOC) consensus statement on Relative Energy Deficiency in Sport (REDs). Br J Sports Med. 2023;57(17):1073–1097. https://doi.org/10.1136/bjsports-2023-106994
35. Roberts LA, Raastad T, Markworth JF, et al. Post-exercise cold water immersion attenuates acute anabolic signalling and long-term adaptations in muscle to strength training. J Physiol. 2015;593(18):4285–4301. https://doi.org/10.1113/JP270570
36. Fyfe JJ, Broatch JR, Trewin AJ, et al. Cold water immersion attenuates anabolic signaling and skeletal muscle fiber hypertrophy, but not strength gain, following whole-body resistance training. J Appl Physiol. 2019;127(5):1403–1418. https://doi.org/10.1152/japplphysiol.00127.2019
37. Pelliccia A, Sharma S, Gati S, et al. 2020 ESC Guidelines on sports cardiology and exercise in patients with cardiovascular disease. Eur Heart J. 2021;42(1):17–96. https://doi.org/10.1093/eurheartj/ehaa605

Support · Privacy Policy · Terms of Use