Wearable-and-camera-based motion tracking has eclipsed clipboard-based observation as the primary objective measure of patient progress. This chapter walks IMU-based wearables, optical motion capture (Vicon, OptiTrack, Xsens), markerless ML tracking, and the data-management envelope under HIPAA + GDPR + per-jurisdiction privacy law.
9-axis IMU sensors (3-axis accelerometer + 3-axis gyroscope + 3-axis magnetometer) at 100–500 Hz on body landmarks deliver per-segment angular kinematics via per-segment orientation estimation (Madgwick + Vaidyanathan 2010 quaternion filter, Mahony + Hamel + Pflimlin 2008, EKF / UKF variants). Joint-angle reconstruction follows ISB recommendations for joint-coordinate-system definition (Wu + Cavanagh 1995, Wu + Siegler 2002 ankle/foot, Wu + van der Helm 2005 shoulder).
Vicon, OptiTrack, Qualisys, Motion Analysis, BTS Bioengineering use retroreflective markers and high-speed infrared cameras (typical 100–250 Hz) for sub-millimetre kinematic accuracy in a calibrated capture volume. Marker placement follows the Plug-in Gait, Cleveland Clinic, or IORgait marker-set conventions. Force plates (AMTI, Bertec, Kistler per ISO 14945) integrate ground-reaction-force signals into the kinematic stream for inverse-dynamics computation.
OpenPose, MediaPipe, AlphaPose, Movement Lab, and proprietary clinical platforms (Theia Markerless, Kinatrax, Captury) deliver markerless 2D + 3D human-pose estimation from monocular or multi-camera RGB video. Accuracy is acceptable for screening + telerehab; gold-standard biomechanics still favours marker-based capture. Per-platform validity must be demonstrated against marker-based ground truth in the patient population of interest before clinical deployment.
The clinician dashboard surfaces per-session repetition-count, per-task quality score, per-session aggregate dose, and trend across sessions. Best-in-class dashboards layer in normative-population comparisons + minimal-clinically-important-difference markers. The dashboard is part of the medical-device software envelope per IEC 62304 — software-of-unknown-provenance (SOUP) components require risk-classified evaluation.
Patient kinematic data is sensitive personal health information per HIPAA + GDPR + LGPD + PIPEDA + per-jurisdiction privacy law. Pseudonymisation (HIPAA Safe Harbor §164.514(b)) before non-clinical analytics, encryption at rest (AES-256-GCM per FIPS 197 + NIST SP 800-38D), encryption in transit (TLS 1.3 per RFC 8446), per-clinician role-based access control, and per-record audit trail are baseline. Cross-jurisdiction data flow follows SCC / IDTA / per-jurisdiction transfer mechanism. Cohort analytics use only de-identified or pseudonymised cohorts; the per-patient consent envelope MUST be in force before any per-patient data leaves the per-clinical-site envelope.