Recorded / device-reported
A value supplied to SomataQuest rather than derived by SomataQuest, such as a device-reported heart rate or sleep duration, or a weight from a workout log.
SomataQuest is built around a simple idea: no single number tells you how recovered, fit, or ready you are. Training load, sleep, heart-rate trends, strength work, and workout history each tell you something different. SomataQuest brings those signals together while keeping clear which values were reported by a device or log, which SomataQuest calculated, and which are only estimates.
These papers are sources we used to justify specific choices. They do not endorse SomataQuest, and they do not validate the app's science. SomataQuest provides informational fitness, training, and recovery context. It is not medical advice, diagnosis, or treatment, and it is not a medical device. Scores are fitness and wellness tools, not determinations that exercise is safe.
Recorded, calculated, and estimated
They are not equivalent evidence.
Calculated by SomataQuest
Values SomataQuest derives from those recorded values using a defined method, such as TRIMP or load-volume.
Estimated
Useful quantities that cannot be measured directly from the available data and therefore depend on assumptions or proxies.
Trends matter more than isolated readings
A single HRV reading, resting heart rate, workout, or night of sleep is not a verdict. SomataQuest compares you with your own recent pattern.
No single metric defines readiness
Load, sleep, heart-rate measures, and workout history are shown together. None of them is ground truth for readiness.
Estimates remain estimates
When a quantity cannot be measured directly, the app labels the proxy, keeps it separate, and does not silently upgrade it into a measurement.
Recovery and heart-rate variability
HRV is highly individual, so SomataQuest cares much more about your own baseline and trend than whether your number is “high” or “low” compared with someone else. It is one recovery signal among several, not a green or red light for training.
Resting heart rate sits beside it. Training, sleep, illness, and stress can move both. We used Buchheit, 2014, to justify following your own trend instead of a universal cutoff.
Sleep
Duration and consistency are the sleep signals worth trusting first. We used Walsh et al., 2021, to justify treating adequate, individualized sleep—and how regular it is—as recovery context.
A watch estimates sleep. Stage breakdowns are not polysomnography, so SomataQuest does not treat a small stage-to-stage change as a harder fact than how long you slept.
Training load
Training load describes how much stress has accumulated. SomataQuest can show duration, intensity, a heart-rate-derived load such as TRIMP, frequency, and intensity distribution—especially when recent work differs from your own history.
ACWR compares recent training load with a longer personal history. SomataQuest uses it to describe how unusual the current load is for you—not to assign an injury probability. We used Bourdon et al., 2017, to justify that descriptive use.
Strength training and volume
There is no single correct number for resistance-training volume. SomataQuest therefore keeps hard sets, reps, external load, and load-volume separate rather than pretending they measure the same thing.
Hard-set count is one useful hypertrophy-oriented volume descriptor. Load-volume is a different description of the same training. Frequency and muscle-group exposure stay visible too, instead of being folded into one stimulus score. We used Hammert et al., 2024, Baz-Valle et al., 2021, and Currier et al., 2026, to justify keeping those quantities distinct and not treating a load total as a hypertrophic dose.
Estimated strength and related exercises
Comparisons between related lifts need an estimate. SomataQuest labels it, lets you replace it, does not merge the two exercise histories, and does not convert total workout volume from one lift to the other.
An optional starting estimate may treat combined dumbbell bench 1RM as about 17% lower than barbell bench 1RM (factor 0.83). That figure comes from a small sample of 12 resistance-trained young men. It is a low-confidence starting estimate, not a universal conversion factor, and it can be replaced in Settings.
Rep-based estimates stay inside a 1–10 repetition window (Reynolds et al., 2006). Above that, prediction gets less reliable, so the result stays an estimated 1RM.
Bodyweight exercise estimates
Some exercises have no logged external load even though you are moving a substantial fraction of your body. SomataQuest can optionally estimate that fraction for standard and knee push-ups (Suprak et al., 2011). The estimate stays labeled and separate from measured external-load volume.
Mindfulness and paced breathing
Slow breathing can change heart-rate variability through respiratory and autonomic pathways (Lehrer & Gevirtz, 2014). SomataQuest can show the heart-rate response during a breathing session as feedback on that session. It is not a clinical measure of autonomic function.
How AI fits in
AI sits downstream of the recorded values and calculations, not in place of them. SomataQuest first builds the training and recovery data itself, including the provenance and limitations of each metric.
The model is not measuring your recovery and it does not get to turn an uncertain wearable estimate into physiological ground truth. Its job is interpretation: connect the supplied signals, explain plausible relationships, and make uncertainty visible.
Coach Today is a separate path: an on-device recommendation with visible reasons. Where AI interpretation is available, it receives structured context assembled by SomataQuest rather than directly generating the underlying measurements or calculations.
AI-generated interpretation is not itself a physiological measurement or clinically validated assessment.
What these metrics cannot tell you
The sections above already reflect these limits. They are collected here so they are not buried in legal copy.
Wearable measurement error
Heart rate, HRV, energy expenditure, sleep, and other wearable-derived metrics can contain measurement error and vary by device.
Sleep-stage uncertainty
Consumer devices estimate sleep stages and should not be interpreted as equivalent to polysomnography.
Training load is descriptive
Load metrics summarize training exposure. They do not provide deterministic predictions of injury, overtraining, or adaptation.
Individual variation
Physiological responses vary substantially between individuals. Personal trends generally matter more than population averages.
Incomplete data
If workouts, sleep, or wearable measurements are missing, SomataQuest only interprets the information available.
Not medical advice
SomataQuest provides informational fitness, training, and recovery context. It is not medical advice, diagnosis, or treatment, and it is not a medical device. See the terms.
Scientific sources
Citing a paper means we used it to justify a choice. It does not mean the authors endorse SomataQuest, or that the paper validates the app.
Monitoring training status with HR measures: do all roads lead to Rome?
Buchheit, 2014
We used this to justify emphasizing your own HR and HRV trend instead of a universal cutoff.
Sleep and the athlete: narrative review and 2021 expert consensus recommendations
Walsh et al., 2021
We used this to justify treating adequate, individualized sleep duration and consistency as recovery context, and not treating wearable sleep stages as laboratory measurements.
Monitoring Athlete Training Loads: Consensus Statement
Bourdon et al., 2017
We used this to justify describing training load as exposure. We do not use those metrics to predict injury.
Heart rate variability biofeedback: how and why does it work?
Lehrer & Gevirtz, 2014
We used this to justify showing a heart-rate response during paced breathing as session feedback, not as a clinical autonomic test.
Progression of total training volume in resistance training studies and its application to skeletal muscle growth
Hammert et al., 2024
We used this to justify keeping hard-set counts and load-volume separate. They describe different things.
ACSM Position Stand. Resistance Training Prescription for Muscle Function, Hypertrophy, and Physical Performance in Healthy Adults: An Overview of Reviews
Currier et al., 2026
We used this to justify not treating a load total as a direct measure of hypertrophic stimulus.
Total Number of Sets as a Training Volume Quantification Method for Muscle Hypertrophy: A Systematic Review
Baz-Valle et al., 2021
We used this to justify showing hard-set count as a hypertrophy-oriented volume descriptor alongside load-volume, not instead of it.
The effect of position on the percentage of body mass supported during traditional and modified push-up variants
Suprak et al., 2011
We used this to justify optional estimated body-mass fractions for standard and knee push-ups. Those estimates stay labeled and separate from measured external-load volume.
A comparison of muscle activity and 1-RM strength of three chest-press exercises with different stability requirements
Saeterbakken et al., 2011
We used this to justify the optional 0.83 dumbbell-to-barbell bench starting estimate. The study included only 12 trained young men, so the conversion is labeled low-confidence and you can replace it. It does not convert volume or merge histories.
Prediction of one repetition maximum strength from multiple repetition maximum testing and anthropometry
Reynolds et al., 2006
We used this to justify keeping related-lift strength estimates inside a 1–10 rep window and labeling them as estimated 1RM. Prediction from higher-rep sets is less reliable.