VR for Physical Therapy & Movement ReVRhabilitation

VR for Physical Therapy & Movement Rehabilitation

Movement rehabilitation is one of the largest real-world applications of VR in clinical settings, but it has been underrepresented on this wiki. This page covers how VR systems — particularly low-cost standalone headsets like the Meta Quest 3 — can be used to capture, analyze, and review patient movement for physical therapy, sports medicine, and post-injury rehabilitation contexts.

Why VR for movement rehab?

Traditional movement assessment in PT clinics relies on either trained-eye observation (fast but not quantitative) or marker-based motion capture in research labs (accurate but expensive and immobile). 2D smartphone video sits between the two: cheap and ubiquitous, but limited to a single viewing angle and difficult to measure precisely.

VR with inside-out body tracking offers a different tradeoff:

Markerless — no setup time per patient; no skin markers, no calibration session

Portable — a $500 headset replaces a multi-camera lab; usable in any clinic room

Quantitative — joint trajectories are sampled directly per frame, not inferred from a 2D viewpoint

Volumetric review — the same recording can be inspected from any angle, with per-frame overlays for joint angles, depth, asymmetry, and tempo

The tradeoffs: tracking accuracy is meaningfully worse than research-grade mocap, occlusion under clothing or by other limbs corrupts data, and the patient typically has to wear the headset themselves — which changes the clinical workflow and may not be appropriate for every population (older patients, patients with vestibular sensitivity, etc.).

Inside-out body tracking on Quest 3

The Meta Quest 3 provides full-body tracking via the Meta Movement SDK using only the headset's onboard cameras — no external trackers, no Kinect, no body suit. The SDK infers 25+ body joints from the user's perspective.

Key things to know for clinical or assessment use:

The user being tracked must wear the headset themselves

Tracking quality depends heavily on lighting and limb visibility

Lower-body tracking (hips, knees, ankles) is consistently weaker than upper-body — directly relevant for many PT use cases

Joint positions are world-anchored; segment lengths can drift if the SDK retargeter is misconfigured (see Diagnosing Meta Movement SDK Retargeting Failures)

Sample rates of 60–90 Hz are typical, sufficient for clinical-rate movements (squats, lunges, gait at walking speed) but insufficient for high-velocity sports motions

For the technical setup of body tracking and a record-and-replay workflow in Unity, see Meta Movement SDK Body Tracking + Record-and-Replay Workflow in Unity.

Existing VR systems for movement rehab

Several commercial and research systems use VR or related technologies for movement assessment and rehabilitation:

Reflexion Health (Vera) — FDA-cleared physical therapy platform using depth sensors (not headset VR but related lineage)

MIRA Rehab — gamified rehab exercises, screen-based with optional VR delivery

Penumbra REAL Immersive System — VR system targeting upper-extremity rehab, particularly post-stroke

Academic prototypes — research groups have published systems for shoulder rehab, post-op knee recovery, balance assessment, and gait analysis using both Quest-class headsets and tethered VR

Most existing systems either gamify exercise (engagement focus) or assess range of motion (clinical metric focus). Comparatively few focus on 3D review of patient movement by a clinician — the use case PT Detective targets.

Biomechanical metrics in VR-tracked motion

Movement screening in PT typically reduces a complex action to a small number of biomechanical metrics, each with literature-backed thresholds. The set below is what PT Detective implements for the bodyweight squat, drawing on Bishop et al. 2017 (asymmetry indices) and Nae et al. 2017 (squat assessment criteria):

MetricWhat it capturesThresholdSourceSquat depthHip-below-knee at lowest point[your value]Nae 2017Knee valgusInward knee collapse on descent[your value]Nae 2017Forward leanTrunk angle at lowest point[your value]Nae 2017Asymmetry indexL/R difference in depth or timing[your value]Bishop 2017Heel lift / weight shiftCenter-of-pressure shift[your value]Nae 2017Knee flexion ROMPeak knee flexion angle[your value]Nae 2017

Fill in the specific thresholds your project uses; values vary by population (athlete vs. clinical) and protocol.

Each of these metrics is well-suited to volumetric review in VR because the clinician can rotate the skeleton to the optimal viewing angle for that specific metric — something a single fixed 2D camera cannot do.

Case study — PT Detective

PT Detective is a Meta Quest 3 application built for Brown CSCI 1951T (Spring 2026). It pairs a Patient capture scene with a PT Review scene: a patient performs squats while wearing the headset; the body-tracked motion is recorded and persisted; a clinician then reviews the recording in a separate scene as a color-coded 3D skeleton with per-frame metric overlays. The full code walkthrough lives at Spring 2026 Project 2 Complete Walkthrough — PT Detective.

A within-subjects counterbalanced crossover study (n=8) compared VR skeleton review against 2D video review for the same set of squat recordings, including simulated anomalies for each of the metrics above. Findings:

All 8 participants preferred VR over 2D video for review tasks (8/8, binomial p ≈ 0.008)

VR rated higher for asymmetry detection, perceived task success, and reduced frustration

2D rated higher for forward lean detection — a single sagittal-view frame turns out to be a strong cue for trunk angle, which the VR skeleton de-emphasizes by allowing arbitrary rotation

Wilcoxon signed-rank tests on per-metric accuracy did not reach significance for most comparisons; with n=8 the discrete distribution of the test statistic places a structural floor on achievable p-values regardless of effect size

Open questions 

Clinical validity — Quest 3 + Meta Movement SDK tracking accuracy has not been independently validated against gold-standard mocap for clinical decision-making

Sample size — classroom user studies (n=8) are structurally underpowered; nonparametric tests struggle to reach significance even with large observed effect sizes

Forward-lean detection in VR underperformed 2D — open design question whether a sagittal "shadow" overlay or a fixed sagittal camera widget could close the gap

Patient experience — having the patient wear the headset (rather than be filmed) changes the clinical workflow and may exclude vestibularly sensitive populations

Lower-body tracking quality is the weakest part of inside-out body tracking and is unfortunately central to many PT use cases (gait, knee, ankle)

Related pages on this wiki

Meta Movement SDK Body Tracking + Record-and-Replay Workflow in Unity

Diagnosing Meta Movement SDK Retargeting Failures in Unity

VR in Medicine (parent page)

Spring 2026 Project 2 Complete Walkthrough — PT Detective

Sanil Desai (Spring 2026)