Reading drift warnings

Two warning banners can appear on a session after the metrics compute. The tare or pretension warning flags force already rising before the contraction began (which shifts onset and biases the early rate of force development). Post-release rebound warns that the trace did not return to baseline after the release (which usually means the strap migrated under load). This walks what each banner is detecting, what it means clinically, and whether to act on it.

Updated October 15, 2026

The platform inspects every recorded trial for a small set of baseline-quality issues that aren't obvious from looking at the metric tiles. When it finds one, it surfaces an inline warning banner on the session page. The banners describe a property of the recording, not a failure of the platform; they exist so a clinician can decide whether the trial is still usable given the issue, override the onset if needed, or exclude and re-run.

Two warnings come up often enough to be worth understanding in detail.

Tare or pretension warning

What it says. "Possible tare or pretension problem: force was already rising before the contraction began, up [amount] over [ms] before onset…", followed by the likely causes and what to do if the rise is large or repeats. Sessions recorded before October 2026 carry the older wording, "Possible tare drift detected."

What the platform is detecting. A steady rise in force in the stretch just before the contraction proper: between where force first starts to move and where the effort clearly begins. The detector fits a straight line to that stretch and fires when the rise is consistent (the line fits well) and lasts at least 50 ms. A flat or noisy baseline does not trigger it; a slow, steady climb into the effort does.

What it usually means clinically. It is almost always a tare or pretension problem. Three common physical causes:

  1. The strap was adjusted, or the patient eased into it, just before the effort. The load changed after the device was tared, so the tare baseline is stale by the time the contraction starts.
  2. The strap is creeping under static load. Some webbing strap materials elongate slightly under continuous tension, which shows up as a slow rise in the force reading leading into the effort.
  3. The patient is pre-tensioning involuntarily. This is harder to spot at the chart level but produces the same pattern: force climbing gently into the contraction because the patient is gradually loading the strap without realizing it.

How to read the banner.

  • If the rise is small (a few newtons across the stretch before onset), the platform's baseline correction probably absorbed it cleanly. The metrics are usable. The banner is informational.
  • If the rise is large (tens of newtons across the same stretch), the auto-detected onset shifted. The early-window RTD is computed against a baseline that doesn't reflect the patient's actual resting state. Consider overriding the onset or excluding the trial.
  • If the banner appears on every trial in a set, the cause is procedural rather than incidental. Retare between trials, check whether the strap is being adjusted mid-set, or switch to a strap material with less creep.

For why some resting pretension is actually desirable on tension dynamometers (it makes countermovement visible), see taring the dynamometer and what a good baseline looks like. Steady pretension is fine: it is a static offset the baseline correction handles. What this warning flags is pretension that is still changing as the effort starts, which moves the onset and biases the early-window RTD.

Post-release rebound warning

What it says. "Post-release force did not return to baseline."

What the platform is detecting. After the patient releases at the end of the trial, the force trace should return to (approximately) the same level it started at. The detector compares the post-release tail of the signal to the pre-onset baseline and fires the warning when the difference exceeds a threshold.

What it usually means clinically. Two common physical causes:

  1. The strap migrated under load. During the maximum-effort contraction, the strap shifted distally or proximally on the limb, and now the resting load on the strap is different from where it started. The dynamometer is reading a different baseline post-release because the rig geometry has changed.
  2. The strap stretched under load. Less common with modern straps, but webbing or buckle hardware can take a small permanent set after a maximum effort. The next trial in the set starts from a different baseline.

Why it matters even though peak and plateau are computed against the pre-onset baseline. The trial's own metrics are usually fine because the platform's baseline correction subtracts the pre-onset offset. The problem is the next trial in the set, which inherits the new baseline. If you don't retare between trials, the next trial's pre-onset window starts at the wrong level, and the platform's correction can't tell that the level shifted between trials versus during a trial.

How to read the banner.

  • If it appears on a single trial within a set, retare before the next trial and continue. The trial in question is usually still usable; the metrics on it were computed against the right (pre-onset) baseline.
  • If it appears on multiple trials in a set, the strap is slipping under load. Tighten the strap, switch to a different strap or attachment point, or accept that this session is going to need a re-tare between every trial.
  • If the trace post-release sits substantially below baseline (rather than above), the gauge may be in a position where the strap is now pulling against it in compression. Check the rig geometry; this is unusual but happens occasionally with creative strap setups.

When neither banner fires but the baseline still looks wrong

The platform's warning thresholds are tuned to surface clear issues without crying wolf. There's a quiet middle ground where the baseline is slightly off but not enough to trip a banner. Two things to watch for visually on the chart:

  • The pre-onset baseline is noisy (fluttering rather than flat). Usually a patient who hasn't settled or a loose strap. The detector still finds an onset, the metrics still compute, but the baseline-noise envelope is wider than usual and onset placement is less precise than usual, because finding onset means finding where the signal leaves that envelope (Tillin et al. 2013).
  • The post-release trace doesn't quite settle back to pre-onset baseline but the difference is small. Subthreshold strap migration. Retake the trial if you can, accept it if you can't, but be aware that the trial's "baseline" is partly a feature of the rig at this point.

For the broader discussion of when these patterns warrant exclude, see when to exclude a trial.

What to do next

  • A banner is information, not an alarm. Read it, decide, move on. Most banners on a single trial within an otherwise-clean session don't require any action beyond awareness.
  • Repeated banners across a set point at procedure, not at the patient. Retare habits, strap material, or rig setup are the things to change.
  • Tare or pretension warnings and post-release rebound warnings are independent. A trial can fire one, the other, both, or neither. The pre-onset and post-release windows are different parts of the signal and the platform inspects them separately.

References

  • Tillin NA, Pain MTG, Folland JP. Identification of contraction onset during explosive contractions. Response to Thompson et al. J Electromyogr Kinesiol. 2013;23(4):991-994. doi:10.1016/j.jelekin.2013.04.015
Was this helpful?