Data Quality
Judging whether a trial is usable
Every number on the screen rests on decisions about the signal underneath it: where the contraction started, whether the baseline was quiet, whether the device sampled fast enough to resolve the window being measured. Those decisions are usually invisible and occasionally wrong. These articles cover what the platform flags, why each flag exists, and which ones change how you read the result rather than whether you keep it.
Choosing the contraction type, MVIC vs RTD vs submaximal hold
The contraction-intent setting tells the analysis layer which metrics belong on the screen. An RTD or ballistic set has no plateau, so steadiness metrics are correctly hidden. A submaximal hold has no explosive rise, so the RTD windows are hidden. This is how to pick and why the choice changes what you see.
Correcting the auto-detected onset
The platform's onset detection is good but not infallible. When the auto-detected onset is wrong, a single click on the chart sets it where it actually belongs and every downstream metric recomputes from the override. The original index is preserved on the session for audit.
Inspecting and editing trials
How to read a captured trial, move through the trials in a session, and decide what each one needs. This is the map that ties the onset-override and exclude-or-delete workflows together, so you know which tool a given curve calls for before you reach for it.
Reading drift warnings
Two warning banners can appear on a session after the metrics compute. Pre-loaded tare warns that the strap was already loaded before the contraction began (which biases peak and shifts onset). 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.
Sampling rate, RTD reliability, and what the badge actually means
Sample rate matters for RTD and almost nothing else. The platform flags low, moderate, and high RTD reliability based on how many samples actually land inside the 0 to 75 ms window. This walks through what the three tiers look like on real signals, which devices land in which tier, and when a low-confidence badge should change your read.
The steadiness reliability flag
The steadiness badge fires when the platform detects that hardware quantization or transition rate is limiting how much of nRMSE / CV / Yank reflects real motor control versus the device's own resolution floor. Device-agnostic by design, and what to do when it appears.
When to exclude a trial, and when to delete instead
Excluding pulls a trial out of patient-level aggregates but keeps it on the record for audit. Deleting removes it from the session entirely. Here's how to tell which one a given curve calls for, with the common cases that come up in clinic.