Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.uplift.ai/llms.txt

Use this file to discover all available pages before exploring further.

The Uplift Movement Library is the complete list of all activities, movements, events, and metrics in the Uplift ecosystem. Use it to determine:
  • which activities and movements Uplift supports
  • what events and metrics are included
  • how the events and metrics are calculated
  • what the biomechanical metrics mean
Uplift refers to movements via both the general activity and specific movement (such as baseball-hitting, jump-countermovement, or track_and_field-discus). Within each activity-movement page, view the required input dimensions as well as all variables (events & metrics) to describe movement performance.

​
Common Terms

  • Session: A single captured movement or attempt, representing the core unit of data collection. These are tagged with a unique alphanumeric session_id. Sessions captured in a sequence (multiple reps or protocol) will have consecutive session_ids (ABC123_1, ABC123_2, …).
  • Dimension: A labeling or categorization system for Sessions, Metrics, or Events, aiding in organization and filtering. Some dimensions are required for proper processing (such as Handedness: left)
  • Activity: the general activity performed (baseball, jump, stability)
  • Movement: the specific movement performed (pitching, countermovement, y_balance_lower_quarter)
  • Event: A specific point in time within the single movement that marks a critical instant or boundary (ball release, jump landing, bottom)
  • Phase: A distinct segment of the movement within a single session, generally bounded by specific events.
  • Metric: A measured or calculated value that reflects some aspect of the athlete’s movement over the duration of the session
  • Discrete Metric: A single, distilled value extracted from the time-series or raw data within a session, used for quick comparisons or reporting.
Note: For brevity, metrics and discrete metrics are often referred to as “Metrics”

​
Technical Details

Each metric and event page includes a Technical Details section that describes how the variable is defined, stored, and interpreted. The parameters are:
  • Variable Type: The role of this variable in the data model.
    • Discrete Metric — a single numeric value extracted from a session (e.g. peak velocity, angle at an event).
    • Continuous Metric — a time-series signal rather than a single scalar.
    • Flag — a boolean per-trial indicator derived from a threshold rule.
    • Dimension — a categorical label used to organize or filter data (e.g. handedness, arm slot type).
  • Data Type: The storage type of the value.
    • Float — a decimal number (e.g. 523.4 deg/s).
    • Boolean — true or false.
    • String — a text category (e.g. "pelvis-trunk-arm").
  • Units: The measurement unit(s) for the metric. Some metrics list two units:
    • raw — the unit returned by the CSV export and Data API (e.g. meters).
    • display — the unit shown in Uplift reports and AI summaries (e.g. inches). When only one unit is listed, it applies to both raw and display values.
  • Column Name: The exact column name in the StarRocks database view. Use this when querying or referencing data programmatically — a mismatch will cause a “column cannot be resolved” error.
  • Aggregation: How multiple trials within a session are collapsed into a single representative value.
    • mean — arithmetic average across trials (used for most continuous metrics).
    • rate — percentage of trials where the flag fired (used for boolean flags).
    • mode — most frequently occurring value (used for categorical/string metrics).
  • Precision: Number of decimal places shown in the UI (e.g. 0 = integer display, 2 = two decimal places).
  • Unknown Value: The sentinel value stored when a metric could not be measured for a trial (currently -1 for all boolean flags). Filter these out before computing statistics.
  • Measurement / Calculation: What the value physically represents or how it is derived from the raw pose data.
  • Timing: The event at which a discrete metric is sampled (e.g. at ball release, at foot contact).
  • Reference: The coordinate frame or zero point used to interpret the value (e.g. 0° = facing home plate).
  • Threshold: For flag metrics, the condition that must be met to set the flag to true.
  • Typical Range: The range of values seen across the broad athlete population. Values outside this range may indicate a data quality issue or an unusual movement pattern.
  • Optimal Direction: How to interpret whether a higher or lower value is desirable.
    • Higher is better — larger values indicate better performance (e.g. peak velocity).
    • Lower is better — smaller values indicate better performance (e.g. time under load).
    • Middle — moderate values are optimal — a range centred around an ideal value; extremes in either direction are suboptimal. See Optimal Range for the target window.
    • N/A — no universal good/bad direction; interpretation depends on context (e.g. arm slot angle, which varies by pitching style).
  • Optimal Range: For Middle metrics, the target window (min–max) that represents ideal performance. Defined as the 25th–75th percentile range from the normative dataset unless otherwise noted.