for people ages 18-70 (full criteria)
study started
completion around
Principal Investigator
by Richard Souza (ucsf)
Headshot of Richard Souza
Richard Souza



More than two million Americans are currently living with a full or partial limb loss, and an additional 185,000 amputations occur each year. The majority of amputations occur in the lower limbs. There are many potential causes for amputation, but the majority can be attributed to vascular diseases, such as diabetes, traumatic injury, and cancer. For these individuals, prosthetic devices play an important role in restoring mobility and enabling them to participate in everyday activities. However, when learning to use these devices, patients often alter their movement patterns to compensate for pain or discomfort, a decreased ability to feel what their prosthetic limb is doing, and/or a fear of falling. By changing their movement patterns, patients will tend to am their intact leg, which has been shown to lead to long-term joint damage and chronic injury. For perspective, 75% of United States veterans living with amputation are diagnosed with a subsequent disease affecting their muscle, bone, and/or joint health. Therefore, therapy sessions, known as gait retraining, are an integral part of teaching prosthesis users to walk in a safe and efficient manner.

With recent advances in wearable technology, researchers and therapists have begun exploring the use of biofeedback systems to assist with this retraining. In these systems, wearable sensors are used to measure how the patient is moving in real-time, and can provide information on how much time they spend on each leg and how much each joint moves during walking. Biofeedback refers to the process of communicating the information from these sensors back to the patients instruct them whether they need to change their movements. Previous research has shown that these systems have excellent potential for helping patients with physical disabilities improve their quality of motion. However, relatively little research has explored how well individuals with above-knee leg amputations respond to biofeedback during gait retraining. Importantly, the question of whether the new movement patterns taught using biofeedback will persist after training has finished remains unanswered.

Therefore, the primary objective of this research is to determine whether biofeedback is a feasible tool for gait retraining with above-knee prosthesis (including a prosthetic knee, ankle, and foot) users. To answer these questions, forty individuals currently using above-knee prosthetic systems will undergo a single session of biofeedback training. Half of these populations will be from the civilian population, and half will be military veterans. During this training, the biofeedback system will apply short vibrations - similar to those generated by cellphones - to their skin every time that the patient reaches the desired degree of hip rotation during walking. Participants will be instructed to keep increasing their hip motion until they feel a vibration on every step. Before training, they will be instrumented with a wearable motion captures system, pressure sensors embedded in their shoes, and a wearable heart rate monitor. Using these devices, researchers will measure the participants' walking patterns without biofeedback determine their current ability. Once training is complete, their walking patterns will be measured again, first while using the biofeedback system, and then again fifteen minutes and thirty minutes after the biofeedback system has been removed. The data measured during these tests will enable researchers to calculate functional mobility scores that are used to evaluate the quality of a patient's walking, and then compare how these scores change before, during, and after biofeedback training.

The knowledge gained through this research constitutes a critical step towards identifying optimal biofeedback strategies for maximizing patient mobility outcomes. The findings will be essential for the development of gait retraining protocols designed to reduce the incidence of chronic injury, and enable patients to achieve their full mobility potential. Building on these results, the next research phase will be to incorporate biofeedback training into a standard six-week gait retraining protocol to evaluate its long-term effectiveness as a rehabilitation tool. Unlike traditional gait retraining, which requires patients to visit clinics in-person for all sessions, the wearable, automated nature of biofeedback training will allow patients to continue gait training from home. This ability will enable patients to continue training activities between sessions, and ultimately may be able to substitute for some in-person visits. This potential for remote therapy has exciting implications for improved access to care for individuals living long distances from their rehabilitation providers, or those suffering from social anxiety, as well as during global health pandemics where in-person visits are difficult.

Official Title

Quantifying Biofeedback Training and Retention Effects on Functional Outcomes in Above-knee Prosthesis Users


Background. Individuals with above-knee amputation (AKA) experience significant mobility limitations due to a combination of factors, including diminished proprioception in the affected limb, that contribute to gait asymmetry, metabolically inefficient movements, and atypical joint loading patterns that have been associated with chronic secondary injuries, such as osteoarthritis and degenerative joint disease. Therefore, increasing gait symmetry, improving balance, and increasing walking efficiency, commonly referred to as gait retraining, is an essential component of rehabilitation for individuals with AKA. Traditional gait retraining is typically limited to a therapist providing verbal cues based on subjective observations to instruct the patient to correct his/her movements to achieve a desired outcome. Biofeedback (BFB) has been gaining attention as a potential tool for supplementing traditional gait retraining by using wearable sensors to quantify a patient's motion and communicate instructions back to the user through externally applied stimuli. Recent research in BFB applications has demonstrated its potential as a rehabilitation tool for several types of mobility impairment, but the effectiveness of BFB as a training tool for improving functional outcomes in individuals with above-knee amputations remains unknown. The proposed research will evaluate the effects of BFB training on the functional mobility of individuals with AKA, and aligns with the Orthotics and Prosthetics Function Focus Area.

Objectives & Hypothesis. The overarching goal driving this proof-of-concept clinical trial is to develop best-practices for incorporating BFB training into gait retraining protocols for individuals with AKA to improve functional outcomes and decrease the prevalence of secondary injury. This study tests the hypothesis that vibrotactile BFB will increase gait symmetry, decrease gait deviations, and improve gait efficiency in above-knee prosthesis users immediately following BFB training; moreover, that these improvements will persist for a minimum of thirty minutes after the cessation of BFB stimuli.

Specific Aims. Specific Aim 1 will test whether short-term BFB training on late stance-phase hip extension produces discernible improvements in overall mobility scores and gait efficiency relative to pre-training levels that persist after the cessation of BFB stimuli. Specific Aim 2 will test whether a single BFB training session produces significant decreases kinematic gait deviations and increases limb loading symmetry relative to pre-training levels that persist after the cessation of BFB stimuli. For both specific aims, the training effects of BFB will be evaluated for (1) efficacy of real-time BFB stimuli, and (2) retention training-induced gait changes after BFB is no longer applied.

Study Design. Forty experienced above-knee prosthesis users will be recruited from the Palo Alto VA (n=20) and University of California Orthopaedic Institute (n=20). Participants will all undergo two types of gait training, in randomized order: traditional (CTRL) and biofeedback (BFBK). Prior to each type of training, participants will be instrumented with a wearable motion capture system and heart rate monitor. They will then perform a two-minute-walk-test (2MWT) to establish a baseline suite of functional outcome levels, including the Physiological Cost Index and gait Symmetry Indices. During BFBK training, BFB will be provided by small vibrating motors when participants have achieved the target amount of hip extension. CTRL training will consist of traditional verbal cuing. After each training phase, participants will undergo an additional 2MWT to quantify the effects of each training type on the outcome measures. For the BFBK phase, participants will undergo two additional 2MWT tests without BFB at 15 and 30 minutes after training to measure training retention. Repeated measures analysis of variance will be used conducted to identify differences in each of the outcome measures between the four time points (pre-test, post-test, 15-minutes, 30-minutes).


Amputation of Knee, Back Pain, Osteo Arthritis Knee, biofeedback, amputation, gait, osteoarthritis, Knee Osteoarthritis, Haptic Biofeedback from SageMotion Biofeedback System


You can join if…

Open to people ages 18-70

  • Unilateral limb amputation occurring between the knee and hip joints
  • Minimum of one-year post-amputation and should have at least three-months experience walking with their current prosthetic system.
  • Classified as K2 and K3-level community ambulators will be included in the study.
  • Able to walk a minimum of 25 minutes with minimal rest. For participants with vascular-related amputations, a doctor's note clearing them to participate in the study will be required.
  • Able to understand and speak English.

You CAN'T join if...

  • Previous spinal cord injury, stroke, or traumatic brain injury.
  • Major upper-limb loss, defined as an amputation occurring proximal to the wrist.
  • Previous or current injury affecting the function of their intact limb (e.g. pelvic injury, osteoarthritis, etc.)
  • Significant hip contracture exceeding 15°
  • Body-mass-index exceeding 35


  • University of California San Francisco accepting new patients
    San Francisco California 94143 United States
  • VA Palo Alto Healthcare System accepting new patients
    Palo Alto California 94304 United States

Lead Scientist at University of California Health

  • Richard Souza (ucsf)
    Richard Souza, PT, PhD, is the Vice Chair for Research and Professor in the Department of Physical Therapy and Rehabilitation Science, with joint appointment in the Departments of Radiology and Biomedical Imaging and Orthopaedic Surgery.


accepting new patients
Start Date
Completion Date
San Jose State University
Phase 1 research study
Study Type
Expecting 35 study participants
Last Updated