Excluding Data in Sift

In biomechanics, some trials or traces may not be suitable for analysis and require exclusion. A missed marker, an incorrect force plate assignment, or an unexpected movement during a trial can all introduce noise into the results. Rather than deleting these trials outright, Sift allows users to flag them for exclusion, which keeps the analysis clean while preserving the original data.

Data cleaning is an ongoing process, and exclusions can be revisited at any time as new issues are identified.

Excluding a trace removes it from analysis (and optionally, plots), but does not delete the underlying data. Exclusions can be undone at any time using Ctrl+Z or by right-clicking a workspace and selecting Re-Include Data, and visibility of excluded traces can be toggled back on to review them alongside valid data.

Sift supports three levels of exclusion:

• Individual trace: exclude a single signal cycle only

• All traces at those frames: exclude all queried signals during the same cycle

• Entire trial: exclude all data from the affected trial

Once exclusions have been identified, they can be written back to the original workspace files using the excluded traces option in the Update CMZs Dialog. This is done by adding a “Bad” event using one of three methods:

• Only exclude individual traces: Flags a single signal for a single cycle. A signal-specific event (e.g. “AnkleAngle_BAD”) is written to the workspace, and the group definition is updated to exclude any cycle containing that event.
• Exclude all traces for this side: Flags all queried signals on the affected side (R/L) for that cycle. Useful when an issue (such as a marker dropout) affects all same-side signals.
• Exclude all traces regardless of side: Flags all queried signals for that cycle, regardless of side. Use this when the entire cycle is compromised.

It is also possible to tag entire trials that contain excluded data, making it easy to identify and filter problematic files downstream.

Two display options in the plot are available:

• Hide Excluded Data: removes excluded traces from the plot entirely.
• Show Excluded Data: keeps excluded traces visible in the plot while still omitting them from analysis.

Clean your Data: A step-by-step walkthrough of using Sift to identify and exclude bad trials across a full dataset, using a ground reaction force quality check as an example.