PostureCheck: a camera tool to detect and correct compensatory posture during robot-assisted arm therapy

PostureCheck: A Vision-based Compensatory-posture-detection Tool to Enhance Performance of the BURT Upper-extremity Stroke-therapy

This will test whether a camera-based PostureCheck system helps stroke survivors with moderate-to-severe arm impairment get better results from robot-assisted arm therapy when used in one-on-one versus small-group sessions.

Quick facts

• Phase: NA
• Study type: Interventional
• Enrollment: 42 (estimated)
• Ages: 18 Years to 80 Years
• Sex: All
• Sponsor: Spaulding Rehabilitation Hospital (other)
• Locations: 1 site (Charlestown, Massachusetts)
• Trial ID: NCT04851886 on ClinicalTrials.gov

What this trial studies

This single-blinded randomized clinical trial will recruit stroke survivors at Spaulding Rehabilitation Hospital and randomly assign them to receive robot-assisted upper-limb therapy delivered either one-on-one or in small groups. A camera-based system called PostureCheck uses machine learning to detect compensatory postures during exercises and can provide feedback to patients and therapists. The trial compares motor outcomes, including Fugl-Meyer Assessment of the Upper Extremity scores, between sessions that use PostureCheck feedback and those that do not, with randomization balanced for baseline FMA-UE and proprioceptive deficits. Outcomes will be measured before and after the intervention period to determine whether PostureCheck improves motor gains.

Who should consider this trial

Good fit: Ideal candidates are adults 6 months to 6 years after an ischemic MCA or ACA stroke with moderate-to-severe upper-limb impairment (FMA-UE 15–45) who score >23 on the MMSE and can follow three-step commands.

Not a fit: Patients with hemorrhagic or posterior-circulation strokes, severe visual impairments or hemispatial neglect, advanced musculoskeletal pathology, very high spasticity, or severe aphasia are unlikely to benefit or may be excluded.

Why it matters

Potential benefit: If successful, PostureCheck could help stroke survivors improve arm recovery by identifying and correcting compensatory movements and could make effective group robot-assisted therapy more scalable.

How similar studies have performed: Robot-assisted upper-limb therapy and feedback systems have shown benefit in prior studies, but camera-based, machine-learning compensatory-posture detection like PostureCheck—particularly in group settings—is relatively novel and not yet widely validated.

Eligibility criteria

Inclusion Criteria:

* Have had a history of ischemic cerebrovascular accident (CVA) within middle cerebral artery (MCA) or anterior cerebral artery (ACA) territory
* CVA occurred between six months and six years ago
* Moderate to severe UE impairment, FMA-UE score between 15 and 45
* Mini-Mental State Examination (MMSE) score \>23 and being able to safely follow three-step commands

Exclusion Criteria:

* Advanced musculoskeletal pathology or recent fractures in the impaired UE
* Previous diagnosis of neurological diseases other than stroke
* Severe limitation of passive range of motion or spasticity (Modified Ashworth Scale for spasticity at \<3 for UE muscles)
* Hemorrhagic CVA or involvement of the posterior cerebral artery
* Aphasia sufficient to limit ability to express needs or discomfort verbally or non-verbally
* Impaired visual acuity (Severe visual impairments as assessed by the NIH Stroke Scale Visual Field subscale. (score \>0))
* Severe hemispatial neglect as assessed by the Line Bisection Test (score \>2)
* Severely impaired trunk control (\<4)
* Individuals who present with the following on the intended limb:
* Open wounds
* Fragile skin
* Under contact precautions due to an active infection
* Participation in another therapy focused on recovery of the impaired arm
* Subjects with cardiac pacemakers, electronic pumps or any other implanted medical devices, that are not US-certified
* Any Condition that would prevent safe use of the Burt system; such as proprioceptive deficits that impair the ability to process haptic or visual feedback, or unstable shoulder joint as assessed by physical examination
* Current pregnancy (self-report)
* The concern is strictly towards women who are in the 2nd or 3rd trimester and the eventuality of the arm of the BURT accidentally impacting the woman's belly. The system does have safety features (including a virtual safety wall, to keep the device away from the patient's body) but we are taking this as an extra precaution against any potential harm

Where this trial is running

Charlestown, Massachusetts

• Spaulding Rehabilitation Hospital — Charlestown, Massachusetts, United States (RECRUITING)

Study contacts

• Principal investigator: Paolo Bonato, PhD — Harvard Medical School (HMS and HSDM)
• Study coordinator: Daniel Tabb, PT

Conditions: Stroke