Help and device checks

Open Capture Lab and choose a versioned movement protocol. Use a fixed camera and the protocol's recording plane; keep required landmarks visible. Complete the calibration and quality checks, then record. Inspect replay and missing observations before exporting or sharing. A camera estimate is a projected measurement, not anatomical 3D motion or a force measurement.

Select video input in Capture Lab. Set a supported sample rate and inspect the recording view. The app processes uploaded video samples in sequence and preserves timestamps and unavailable observations. Source video remains on your device; retain its original file under your study's data policy.

Low visibility, motion blur, camera angle, occlusion and incomplete calibration can make measurements unavailable. Missing values are not zero. Review the raw landmark replay and quality flags. Do not replace unavailable observations with invented values or interpret model visibility as an accuracy confidence interval.

Open a trial in Research, review its timeline and observations, add annotations and a review decision, and export CSV or JSON. Local reviewer names are labels. Authenticated institutional reviews are created in the Institution workspace after an authorized upload.

Register a scoped observed-data plan before importing evidence. Supply the independent reference, angle convention, clock anchors, uncertainty, participant/occasion codes and study criteria. Inspect aligned samples, exclusions and confidence intervals. A result applies only to its registered scope; it is not automatic approval of a commercial accuracy claim.

Synthetic examples are explicitly labelled for teaching and software verification. They can demonstrate the workflow and known numerical offsets. They cannot be imported as observed human validation data or used as normative reference values.

Use your institution's configured sign-in. An owner provides a one-time invitation bound to your verified email. The assigned role controls uploads, raw exports and review. If sign-in or an invitation fails, ask your administrator to check directory membership, the invitation and the configured identity provider.

Central uploads require a current matched consent record. Withdrawal removes linked central trials and pending uploads. Device records, source videos, exported reports and historical backups are separate copies: reconcile them against the withdrawal register. Validation Lab includes participant erasure for its local study data and saved reports.

Export encrypted backups from Research and separately from each Validation Lab study. Keep the passphrase in your institution's approved password manager; there is no account recovery key for these archives. Test restoration before replacing a device. Private browsing, clearing site data or a changed app origin can make local records unavailable.

Run the device checks here and download the support report. It contains capability results and a broad browser family, not participant records, camera images, account IDs or URLs. Provide the request ID from an institutional error and a description of the action to your administrator. Review screenshots and other attachments for participant information before sharing.

KINEOS supplies raw estimated landmarks, defined projected variables and study-specific analyses. The shipped release has no genuine paired human accuracy dataset. Force, torque, EMG, diagnosis and injury probability are not inferred by this camera pipeline. Clinical or commercial measurement claims require separate evidence and investigator approval.