Whole Body Sweat Rate Prediction: Indoor Treadmill and Cycle Ergometer Exercise
Participants
n=109 participants in the cycling models (67 development + 42 validation; 112 + 56 = 168 cycle-ergometer trials). Treadmill: 67 development + 63 validation participants (171 + 95 = 266 trials). Fewer female subjects, but their values did not differ from those expected for males.
Protocol
Conditions matched between development and validation: dry-bulb 20-40°C, relative humidity 14-60%, intensity ~40-85% of peak aerobic power
Whole-body sweat rate measured; proprietary models built on the relevant biophysical factors, then validated on the independent group
Inputs: simple parameters where power output is measured or derived (the abstract does not list the full input set)
Key Results
| Metric | Cycle ergometer | Treadmill |
|---|---|---|
| Mean absolute error | 0.01 L/h | 0.02 L/h |
| 95% confidence interval | +0.25 / −0.22 L/h | +0.33 / −0.29 L/h |
| Variance accounted for | 92% | 78% |
| Trials beyond the a priori agreement limit (±2% body mass over 3 h) | 0 | 1 |
Public calculator: https://sweatratecalculator.com
Key Findings
First study to develop and validate whole-body sweat-rate prediction equations for indoor cycle-ergometer and treadmill exercise at moderate-to-high intensity
Accuracy held across temperate-to-hot conditions in the independent validation group
Indoor only: the models were built for ergometer and treadmill exercise, where power is known and there is no wind or sun
Practical Application
In an independent validation group, the cycle-ergometer model predicted whole-body sweat rate with a mean absolute error of 0.01 L/h (95% CI +0.25/−0.22 L/h, 92% of variance explained) across 20-40°C, 14-60% RH and ~40-85% of peak aerobic power; no cycling trial missed the ±2% body-mass-in-3-h agreement limit. For indoor sessions on a smart trainer this gives a trustworthy starting estimate of L/h from power and room conditions.
Inference, not tested: outdoors, airflow, sun and clothing change sweat rate; the published validation is indoor only, so treat an outdoor estimate as a hypothesis to check by weighing.
Not from this paper; the base's own practice: use the calculator's number to set bottles per hour for turbo sessions, then confirm with a before/after weigh-in; do not drink back 100% of predicted loss.
Limitations
From the design as the abstract reports it: prediction-model development and validation; the models are proprietary, and the abstract does not give the equations or the full input set.
Indoor cycle-ergometer and treadmill exercise only, within 20-40°C, 14-60% relative humidity and ~40-85% of peak aerobic power.
Fewer trials with female subjects (their values did not differ from those expected for males).
How this entry was checked and changed
Why this level
regraded 2026-09-27 from B+: test/model validation study
Content correction 2026-09-27
"Validated prediction equations ... during OUTDOOR cycling" -> the study is indoor (cycle ergometer and treadmill), per title and abstract.
"158 cycling trials" -> 112 + 56 = 168 trials (67 + 42 participants). "MAE ~0.02 L/h for cycling" -> 0.01 L/h (0.02 is treadmill). "95% CI ±2% body mass" -> the 95% CI is +0.25/−0.22 L/h; ±2% body mass in 3 h was the a priori agreement limit, exceeded by zero cycling trials.
Input list "power, temperature, humidity, body mass, age" -> the abstract names only "simple input parameters" with power; list removed. Worked example "250 W, 25°C, 60% RH, 82 kg → ~1.3 L/h, 910 mL/h" -> not from the paper; removed.
Journal line "Medicine & Science in Sports & Exercise" -> the DOI is Journal of Applied Physiology. Level B+ -> B (validation study). Sport line adds running (treadmill arm).
2026-09-28: added Limitations from the abstract (Europe PMC/PubMed PMID 39205641).