The research behind your plan

Safety first

Udapt is designed so it can't push anyone harder than their plan, and so people get professional advice where they need it.

• Talk to a professional before starting. During setup Udapt says: "Before you start, check with your doctor or a health professional that exercise and food changes are right for you." The plan summary repeats it, "especially if you're pregnant, have a medical condition, take regular medicine or haven't exercised for a while."
• A health check before any plan is shown. Five yes-or-no questions, modelled on the PAR-Q+ screening form and the American College of Sports Medicine's pre-exercise screening guidance, ask about heart conditions or surgery, chest pain, fainting or dizziness, heart-rate or blood-pressure medicine, and recent surgery or exercise limits. The wording was reviewed by a practising registered nurse.
• Any "yes" and Udapt steps back. It stops suggesting or changing workouts and says: "Your medical team's plan comes first." It only switches back on if the person confirms their doctor or rehab team has cleared them, and that can be withdrawn at any time.
• Plans from a physio or rehab team are recorded, never changed. "My own plan" logs the programme a person's clinicians gave them exactly as written, with no progression, no swaps and no alerts.
• Stop-and-get-help advice on every session of a clinician's plan, from NHS, British Heart Foundation and cardiac rehabilitation guidance: stop for chest pain, dizziness, an irregular heartbeat or uncomfortable breathing, and call the local emergency number for tight or spreading chest pain.
• Heart-rate medicines are respected. Medicines such as beta-blockers change resting heart rate and heart rate variability, so Udapt leaves those signals out of readiness for anyone taking them.
• It only ever makes a day easier. Readiness can reduce a session or suggest rest. It never adds work beyond the plan, and the original session is always one tap away.
• Unwell means rest. With a fever, chest symptoms or whole-body aches, Udapt suggests rest, not even a walk, and eases people back in over a week.
• Teenagers (13–17) get different rules: no weight-loss goals or calorie deficits, up to 3 strength days a week, adolescent calorie equations, effort kept below maximum, and advice to learn lifts with a coach or teacher, show their plan to a parent or carer, and talk to a doctor about any health worries.
• Stop if… before every run, ride and swim. "Stop if you get chest pain, unusual breathlessness, feel dizzy or faint, or your heartbeat races or feels irregular", with the local emergency number. Only people who answered no to every health-check question can hide it, in Settings. People who answered yes to a heart question are pointed to their own team's plan for walking, running or cycling. Stop signs as in the NHS, British Heart Foundation and cardiac rehabilitation guidance above; heart patients' own plans: EAPC position statement (Hansen et al. 2022) and NICE NG185.
• Pain isn't part of the plan. The app says: "Effort is fine; pain isn't. If an exercise hurts, stop and swap it, and talk to a doctor or health professional."

Back to contents

Readiness: how ready you are today

Each morning Udapt gives one of three readiness states: Go (train as planned), Ease off (a lighter session) or Recover (a walk, mobility or rest), and says how many of its checks were in the person's usual range, for example "1 of 2 checks in range". There is no score out of 10: no study gives one. Two checks decide the state, each taken from a published study.

1. Heart signals. Udapt multiplies the person's overnight heart rate variability (HRV) by their resting heart rate and compares the result with their own last 10 readings. If today is below the average of those readings minus one standard deviation, that counts as one step down. Without a resting heart rate, HRV is used on its own.
2. Morning check-in. Four questions: sleep quality, tiredness, muscle soreness and stress, each from 1 to 7 where 4 means "usual for me". If tiredness or soreness is above 5, or the two add up to 10 or more, that counts as one step down. Sleep quality and stress are recorded and shown back, but don't change the score.
3. The result. No steps down is Go. One is Ease off. Two is Recover. The app explains this in its "How your readiness is worked out" section.

Applied Heart-signal rule, the 10-reading reference, and "each check out of range is one step down". Alfonso et al. 2025 (28 experienced cyclists; HRV, resting heart rate and well-being together gave the largest improvements). It builds on Kiviniemi et al. 2007. Studied in endurance training; Udapt applies it to strength training.

Applied Check-in questions, the 1–7 scale and the "above 5, or 10 or more together" cut-off. Nuuttila et al. 2022 (30 recreational runners), who took the cut-off from Hooper et al. 1995.

Direct How people say they feel is a useful signal. Saw et al. 2016, a review of 56 studies: self-reported measures tracked training load at least as well as objective ones.

Direct Sleep is shown next to the readiness state with the expert guidance of at least 7 hours (8–10 for teenagers), but doesn't change it, because no training study used it this way. Watson et al. 2015; Paruthi et al. 2016.

Applied For people who track their cycle, the heart-signal reference uses only days from the same cycle phase once two full cycles are logged, so a normal dip in HRV before a period isn't read as poor recovery. Schmalenberger et al. 2019 (HRV falls in the luteal phase).

The morning reading (since Udapt 1.5)

For people who don't wear their Apple Watch in bed, and only if they turn it on: a 1-minute Breathe session on the Watch when they wake stands in for the overnight reading, and goes through the same heart-signal rule above.

Applied Taken on waking, sitting, breathing at a normal pace, before coffee, food or exercise, and confirmed in the morning check-in; otherwise it doesn't count. Alfonso et al. 2025 measured "first thing in the morning, in a seated position, before any other movement, and in a fasted state"; Altini & Plews 2021 measured "right after waking up" to limit "the effect of other stressors (e.g., caffeine intake or physical activity)". Breathing normally rather than following the animation: Bonneval et al. 2025 "encouraged [participants] to breathe at their own pace"; slow, paced breathing raises HRV while you do it (Laborde et al. 2022, 223 studies).

Applied One minute is long enough, after about a minute sitting still. Esco & Flatt 2014 (ECG in 23 college athletes, lying down): "60 seconds appears to be an acceptable recording time".

Applied Morning readings are compared only with other morning readings, and overnight readings only with overnight ones. On a day with both, the overnight reading is used. Nuuttila et al. 2024 (24 recreational runners): morning and night-time readings correlated in a normal training period, but "their responses to intensive training were not similarly aligned"; the night-time readings "seemed to be sensitive to physical loading".

Applied HRV is worked out as RMSSD from the session's heartbeat timings. Missed, extra and irregular beats are found and corrected with Lipponen & Tarvainen's 2019 method, whose limits adapt to each person's own beat-to-beat variation, and at least 30 beat-to-beat differences are needed. With too few clean beats, Apple's own SDNN reading is used instead, compared only with other SDNN mornings. Lipponen & Tarvainen 2019 (validated on resting ECG, not on a watch or 1-minute readings; the authors founded Kubios, which uses the method); Munoz et al. 2015 (3,387 adults: 30 seconds of beats gave a valid RMSSD); Bourdillon et al. 2022 (changes of 30 ms or less had "negligible influence on RMSSD", so the correction’s limits are never below 30 ms).

Udapt’s choice Which reading counts: the first Breathe session on the Watch that day, started before noon and before any workout. The research says "on waking" but gives no hour, and without sleep data Udapt can't see when someone woke. HRV readings the Watch takes by itself during the day are never used: they come at random times.

Direct Its limits, which the app states: the readiness studies used chest straps or rings, not Apple Watch; Apple Watch HRV during Breathe sessions was about 30% off a chest strap or ECG on average; and no study has guided training with Breathe-session readings. O'Grady et al. 2024 (39 adults, readings on waking: average error 28.9%); Bonneval et al. 2025 (78 adults: 31.3% at rest).

Back to contents

How training adapts

Direct Programmes train every major muscle at least twice a week. Schoenfeld, Ogborn & Krieger 2016: "the major muscle groups should be trained at least twice a week to maximize muscle growth."

Direct Sets and wide rep ranges: more weekly sets give more gains, and muscle builds across a wide range of reps when sets end close to failure. Schoenfeld et al. 2017; Schoenfeld et al. 2021.

Direct Effort: stop each set with about 2 reps left. Training to failure isn't needed. ACSM position stand 2026 (Currier et al.): "a target of 2–3 repetitions in reserve"; Refalo et al. 2023.

Direct Progression: add reps first, then weight. Weight goes up by about 5% for big leg lifts and 2.5% for everything else, once every set reaches the top of the rep range. Plotkin et al. 2022 (adding reps or adding weight both work); ACSM 2009: "2-10% increase in load" when the lifter can do one or two reps more than the target.

Direct Missing the bottom of the range two sessions running: the person picks a new weight they could lift with about 2 reps to spare, rather than Udapt guessing a cut. ACSM 2026 (2–3 reps in reserve). No study sets the size of a step down, so Udapt doesn't invent one.

Direct Rest: 3 minutes for big lifts, 2 for smaller ones. Schoenfeld et al. 2016 (3-minute rest beat 1 minute); ACSM (Garber et al. 2011): "Rest intervals of 2–3 min between each set of repetitions are effective."

Direct Warm-up: one set of 10 at half the working weight before the first exercise. The warm-up used by Ribeiro et al. 2014; Iversen et al. 2021 advise limiting the specific warm-up to the first exercise. Ribeiro found warming up made no difference to reps, so the evidence is thin.

Applied Ease off: about a third fewer sets, stop with about 3 reps left, no weight increase that day. ACSM 2026 (2–3 reps in reserve); Bickel et al. 2011 (a third of the training dose kept muscle size in young adults); Rogerson et al. 2024 (volume cuts of about 30–70% before competition).

Applied Recover: an easy walk, mobility or rest; the person chooses, and the planned session stays available. Kiviniemi et al. 2007 ("low-intensity training or rest" on a low-HRV day); ACSM (Garber 2011) on flexibility work.

Applied A lighter week, offered on a recovery day and taken only if the person chooses: fewer sets, the same number of sessions, for 7 days. Bell et al. 2023 (expert consensus: a deload "could occur when athletes feel physically and mentally fatigued"); Rogerson et al. 2024 (246 lifters; most kept the same sessions and deloaded for about a week).

Applied Resistance bands: people can train with bands and bodyweight only, with the same programmes and the same twice-a-week rule. Bands have no weight in kg, so progress is by adding reps, in ranges of 10–15 and 12–20. The back is trained with band reverse flies and pull-aparts, which work the upper back between the shoulder blades but not the lats. Lopes et al. 2019 (meta-analysis of 8 trials, ages 15–88): "Elastic resistance training is able to promote similar strength gains to conventional resistance training"; ACSM 2026 (8–20 reps per set); Garber et al. 2011 ("10–15 repetitions" for people starting out); Fukunaga et al. 2022 (band pull-apart muscle activity, 10 healthy men).

Direct Exercise how-to text comes from the public-domain free-exercise-db, corrected where it was wrong: knees may travel past the toes in a squat if comfortable; the band good morning loops the band across the back of the shoulders, never around the neck. Fry et al. 2003; free-exercise-db's own barbell good morning ("across the rear of your shoulders").

Direct Joint protection: a person can flag a joint (for example a knee) and Udapt swaps to exercises that don't load it. Which exercise loads which joint follows measured joint loads where studies exist: squats load the "ankle, knee, hip and spinal joints"; the bent-over row gives "the largest lumbar spine load" and the inverted row "the lowest". Schoenfeld 2010; Escamilla 1998 and 2001; Cholewicki 1991; Fenwick et al. 2009; Fees et al. 1998. Pulls, flyes, curls and raises follow the joints they move.

Direct Strength charts estimate a one-rep max with the Epley formula from sets of 10 reps or fewer, for charts only, never to set a weight. Nickerson et al. 2020 (lowest error of four methods against measured one-rep maxes); McNair et al. 2011; Reynolds et al. 2006 ("no more than 10 repetitions").

Back to contents

Running, cycling and swimming

Udapt's own cardio plans are offered only to people whose health check is clear. Each session runs on iPhone or Apple Watch with a timer, spoken cues and haptics, and is saved to Apple Health. Runs, rides and swims that other apps and devices save to Apple Health are counted too.

Direct Run/walk follows the NHS Couch to 5K structure as published: nine stages of three sessions, a five-minute brisk walk before and after, ending with a 30-minute run. Any week can be repeated: "If you're not quite feeling ready, it's totally okay to repeat a week." NHS Couch to 5K plan. Udapt uses its own words and no NHS branding.

Direct After a break, run/walk carries on where the person left off, and "Repeat a week" starts this week or any earlier one again from its first run, whenever the person chooses. It never moves anyone back on its own, and no number of days off triggers anything, because no source gives one. NHS Couch to 5K: "You can always come back to the plan at any time"; "if you need to repeat any of the runs or weeks, that's totally okay".

Applied Running towards a goal: weekly running time or distance never rises more than 30% over two weeks. Nielsen et al. 2014 (exploratory cohort of novice runners). The popular "10% rule" isn't used: in a trial of 532 novice runners it made no difference to injuries (20.8% vs 20.3%; Buist et al. 2008).

Direct Easy effort is judged by the talk test, "you could talk in full sentences". No heart-rate zones by default. EAPC position statement (Hansen et al. 2022) on the talk test as a practical guide; Reed & Pipe 2014 (comfortable speech is possible below the ventilatory threshold).

Applied Experienced runners and riders with three or more days a week get one steady session; everything else stays easy. Rosenblat et al. 2025 (individual-data network meta-analysis: recreational athletes "may improve more" with mostly easy and some moderate training); Reed & Pipe 2014 for what "steady" means.

Direct Cycling: rides of the length the person chooses, 30 minutes suggested (20 for beginners), with no automatic weekly increase because no reputable source gives one for new cyclists. ACSM (Garber et al. 2011): "30–60 min·d−1 … of purposeful moderate exercise".

Direct Swimming: starting sessions of 300 m, 950 m or 2,000 m by level, in the pool or open water. People who can't yet swim are pointed to adult lessons. Swim England fitness training plan and Adult Swimming Framework.

Applied Readiness: on an Ease off day the last session is repeated with no step up; on a Recover day, an easy walk or rest. NHS Couch to 5K (repeating is fine); Kiviniemi et al. 2007 ("low-intensity training or rest").

Direct After each session: "How hard was that session?" from 0 to 10, then the talk test, then whether to repeat the week. Not being able to talk is vigorous effort, so repeating is suggested after a "No", and it's always the person's choice. Haddad et al. 2017 (session rating of effort, after Foster); NHS physical activity guidance on vigorous effort.

Back to contents

Levels and race goals

Direct Three levels for strength, running, cycling and swimming: Beginner, Intermediate and Experienced. Beginners do two sets, 8–12 reps on the big lifts, and dumbbell, kettlebell or machine versions of the barbell lifts. ACSM 2009 (novice "8-12 repetition maximum"); ACSM 2026 (experience has "a minimal impact on strength and hypertrophy outcomes", "at least two sets per exercise", and machines versus free weights don't change strength).

Applied Moving up is offered, never automatic: Intermediate strength after a session in 20 of the last 26 weeks; running after the run/walk plan; swimming after swimming the next level's distance. "Not now" puts it off for 30 days. ACSM 2009 (intermediate: "approximately 6 months of consistent" training).

Applied Race goals for 5K, 10K, half marathon or marathon: the last two weeks before race day are lighter (about 30% less volume, then 60% less in the final week), with the same effort and the same days. Bosquet et al. 2007 (meta-analysis of 27 studies: a two-week taper with volume down 41–60%). Predicted finish times use Riegel's formula up to the half marathon, and none is given for the marathon (Vickers & Vertosick 2016, 2,303 runners).

Direct Race pace (since Udapt 1.5), for race day only: the predicted finish time divided by the race distance, from 5K to half marathon, shown to the nearest 5 seconds per km or mile. None for the marathon. Vickers & Vertosick 2016: Riegel's formula was "well-calibrated for races up to a half-marathon", for "more accurate race time predictions for better pacing".

Applied The recent race must be from the past year; with an older one, Udapt asks for a newer race and shows no pace. Cheuvront et al. 2026 (19 runners) "required that all race times provided were achieved within the past year", following Daniels: estimates are most reliable when "achieved during periods of similar training status".

Direct No pace targets for training runs, at any level: easy and steady runs stay by the talk test, which moves with the person's fitness, and readiness never changes race pace. Foster et al. 2008 (talk-test intensity changed with the ventilatory threshold); Woltmann et al. 2015: "the TT can be used as a primary method to control exercise training intensity". No study has validated an easy pace worked out from a race for recreational runners (Scudamore et al. 2018 tested only interval and threshold paces, in 9 recreational runners).

Applied On race day: "Aim for about [your race pace]. On a hot day, expect to be slower." After every run: "Hills and heat make the same pace harder. That's normal: go by the talk test." Words only; nothing in the plan changes. Ely et al. 2007 (marathon results slowed as heat rose, more for slower runners; applied to 5K–half marathon); Minetti et al. 2002 (running uphill costs more energy); Wingo et al. 2020 (in heat, the same pace becomes harder as heart rate drifts up); Racinais et al. 2015 (consensus: endurance events in the heat are run "at a lower work rate").

Applied Fitness training, for adults already at Intermediate, built around the published rules of a fitness race: Udapt's runs on cardio days, the person's strength programme unchanged, and one station session a week, half the race in its own format (four 1 km steady runs, each followed by a station at the rulebook's amount and the chosen division's weight). The race's rules set a minimum age of 16; Udapt offers it to adults only. HYROX Rulebook (Singles 26/27); Brandt et al. 2025 (the race is mostly running; 11 athletes); Chittenden et al. 2026 (injuries mostly overuse, knees and lower legs; 89 athletes). No training trial for this kind of race exists, so station practice follows the ACSM principle of training specifically.

Direct Medicine ball: with a medicine ball at home or in the hotel gym, each session can end with one power exercise, 2 sets of 3–6 light, fast reps. ACSM 2009 (power training: 1–3 sets of 3–6 reps with light loads, alongside strength training).

Applied Training anywhere: each session can be at the gym, at home, with resistance bands, on the floor or in a hotel gym, and every version still trains each major muscle at least twice a week. Schoenfeld, Ogborn & Krieger 2016; Lopes et al. 2019 for bands.

Back to contents

Food targets

Direct Starting calorie estimate: resting energy from the Mifflin–St Jeor equation, multiplied by an activity factor from 1.45 to 2.0. Mifflin et al. 1990; Frankenfield et al. 2005 (the most reliable of the common equations); FAO/WHO/UNU 2004 activity bands.

Direct The estimate then learns from the person's own weigh-ins and food logs, but only after 31 days with a weigh-in and a complete food log every day. Hall & Chow 2011 (daily weights over more than 28 days); Basiotis et al. 1987 (about 31 days of records to estimate one person's usual intake within 10%).

Direct Weight loss is capped at 1% of body weight a week (0.5% by default), and gain at 0.5% (0.25% by default). Helms et al. 2014: losses of "approximately 0.5 to 1%/wk to maximize muscle retention"; Garthe et al. 2011; Iraki et al. 2019: "a target weight gain of ~0.25-0.5% of bodyweight/week".

Direct Protein: 1.6 g per kg of body weight by default, adjustable from 1.2 to 2.0, never reduced. Morton et al. 2018 (no further gains above about 1.62 g/kg/day); ISSN position stand (Jäger et al. 2017): 1.4–2.0 g/kg/day "is sufficient for most exercising individuals"; GLP-1 advisory (Mozaffarian et al. 2025): "prolonged intake at or above 2 g/kg/day should be avoided".

Applied For people with a BMI of 30 or more, protein is set from the weight at a BMI of 25. Mozaffarian et al. 2025: "the use of actual weight can significantly overestimate protein requirements"; a corrected weight is one of the options it names, with no consensus yet.

Direct Fat at 25% of calories, within the 20–35% range, and never below 0.6 g per kg. Institute of Medicine 2005 (Acceptable Macronutrient Distribution Range); Iraki et al. 2019 (fat "0.5-1.5 g/kg/day").

Direct The same target every day. No study gives a figure for eating more on a weight-training day, so Udapt doesn't shift calories between training and rest days.

Direct Calories never drop below the higher of resting energy or 1,200 kcal (women) / 1,500 kcal (men). If a goal would need less, the pace slows instead and the app says why. NHLBI Practical Guide 2000; AHA/ACC/TOS guideline (Jensen et al. 2014).

Udapt's choice For people who prefer not to give their sex, the starting estimate uses the average of the published male and female equations.

Applied Known limits we disclose: people tend to under-record what they eat, and the standard 7,700 kcal per kg figure (Wishnofsky 1958) overstates the deficit in the first weeks of a diet and for leaner people. Burrows et al. 2019 (59 studies); Thomas et al. 2014; Hall 2008.

Direct Food data comes from official national tables: USDA FoodData Central (US), CoFID 2021 (UK) and CIQUAL 2020 (France), plus Open Food Facts for barcodes. Photo and voice logging use Apple's on-device model only to name the foods. Every nutrition figure comes from those databases, never from the AI.

Back to contents

Life stages and days off

Menstrual cycle

Direct Udapt does not prescribe "cycle-phase workouts": research finds the cycle has at most a small, individual effect on performance. McNulty et al. 2020; Colenso-Semple et al. 2023.

Direct No extra calories before a period. A review found resting energy use rises slightly in that phase, but not significantly in studies since 2000, and gives no calorie figure. Benton et al. 2020.

Direct For people who track their cycle, the pre-period phase is taken as the 12 days before the next expected period. Bull et al. 2019 (612,613 cycles; average 12.4 days).

GLP-1 medicines (such as semaglutide or tirzepatide)

Direct An optional mode: the protein default (1.6 g/kg) sits within the 1.2–1.6 g/kg proposed during weight loss, and "Queasy today" or "Not hungry today" brings up the advisory's advice: small, protein-first meals every 3–4 hours and plenty to drink, and makes a Go day's session lighter (as Ease off), because eating less than usual leaves less energy for training. Anyone feeling unwell can mark the day as unwell. Low energy availability: Mountjoy et al. 2023 (IOC consensus); registered nurse review. Joint advisory of four US obesity and nutrition societies (Mozaffarian et al. 2025); side effects mostly mild and short-lived (Wharton et al. 2022).

Direct Udapt never gives dose, timing or missed-dose advice, and never comments on starting or stopping the medicine.

Teenagers (13–17)

Direct No weight-loss or weight-gain goals: the target is the estimated need for age and sex, which already allows for growth, and never above the recommended intake. NICE NG246 (2025, updated 2026); SACN 2011; Golden et al. 2016 (American Academy of Pediatrics).

Direct Adolescent energy equations and growth-adjusted activity levels. Henry 2005; SACN 2011.

Direct Up to 3 strength days a week with a rest day between (runs, rides and swims are separate), at least 6 reps a set, effort below maximum, weight steps of 2.5–5% (smaller than the 5–10% the NSCA suggests), and advice to learn lifts with qualified instruction. NSCA youth position statement (Faigenbaum et al. 2009).

Applied Under 14: resistance bands and bodyweight only; weights from 14. The age is a safety position from our registered nurse's review, not a UK rule: no UK minimum age for lifting was found, and the NSCA says "there is no minimum age requirement". The method follows the NSCA: bands and bodyweight "have been used by children and adolescents", and "it is reasonable to start with relatively simple exercises and gradually progress". NSCA youth position statement (Faigenbaum et al. 2009).

Applied The learning calorie estimate is off for under-18s: the method was built on adults, and growth would read as overeating. The top activity level (1.93) is SACN's 90th centile with its growth adjustment. Hall & Chow 2011; SACN 2011.

Perimenopause and menopause

Direct An optional mode. Strength programmes stay the same (two to four days meet "two or three days" a week), with one back-extension exercise a week, and no sit-ups, crunches or loaded spinal bending are ever added. UK consensus statement on exercise for osteoporosis (Brooke-Wavell et al. 2022, Box 1).

Direct An optional daily Bone builder follows the Royal Osteoporosis Society's three impact levels in its own words. Everyone starts at Level 1 and moves up only when they choose. Anyone who has broken a bone in their spine, or more than one bone in a minor fall, stays at low impact and is advised to speak to a health professional first. Royal Osteoporosis Society, "Impact exercise to help make your bones stronger"; Brooke-Wavell et al. 2022.

Direct Two optional check-in taps, hot flushes or night sweats and achy joints, are recorded and shown back as the person's own pattern; they don't change readiness. Udapt never comments on hormones, HRT or any medicine.

Unwell and travelling days

Direct Unwell: rest. A gentle walk is suggested only for head-and-throat symptoms, never with a fever, chest symptoms or whole-body aches. Udapt doesn't teach the "neck check" rule, which has no evidence behind it. Ruuskanen et al. 2022.

Applied Coming back: three lighter days, then four eased days, then normal. Jooste et al. 2026 (athletes returned to any training after a median of 3.5 days and to full training after 8).

Applied Travelling: hotel-room exercises at lighter-week volume. Rogerson et al. 2024 for the volume.

Back to contents

What we don't claim

• Udapt isn't proven to build extra muscle. Two small trials that used HRV to decide when people lifted found the same gains as a fixed schedule (de Oliveira et al. 2019; Bittencourt et al. 2024). Readiness is a way to manage tiredness.
• The readiness studies are small. 28 cyclists (Alfonso 2025, with many drop-outs and no fixed-plan group) and 30 runners (Nuuttila 2022), both in endurance sport. No readiness measure, ours included, has an agreed standard to be tested against (Schaffarczyk & Sperlich 2026).
• Most of the evidence comes from athletes, not the everyday adults who use Udapt.
• Apple Watch HRV is less precise than a chest strap (O'Grady et al. 2024), so Udapt compares people only with their own readings.
• The morning reading is a weaker measurement. Breathe-session readings were about 30% off a chest strap or ECG on average (O'Grady et al. 2024; Bonneval et al. 2025), and no study has guided training with them.
• Race pace is an estimate for race day, from one recent race, not a target for training.
• Udapt is not a medical device and not a rehabilitation programme. It doesn't diagnose, treat or monitor any condition.

Back to contents

All sources

Every source Udapt's rules rely on, what it supports in the app, and how directly. Links open the published paper or guideline.

iSafeNet · udapt.isafenet.app · Udapt is a wellness app. It doesn't diagnose, treat or monitor any medical condition and isn't a substitute for advice from a doctor, physiotherapist or registered dietitian.

Back to contents

How we check the evidence

Every rule was checked against the original research, and where the research didn’t support a rule, we changed the rule or removed it. Udapt 1.3’s rules were re-checked in October 2026: each paper against its published abstract, and each rule against what the app actually does. Udapt 1.5’s new rules, the morning reading and race pace, were checked against their sources in October 2026, before the update went to Apple.

Last reviewed 6 October 2026. Spotted something we’ve got wrong? Email info@isafenet.app.