Biofeedback in rehabilitation

The biofeedback measurements which are frequently used in physical rehabilitation can be categorised as being either physiological or biomechanical (Figure 1). The physiological systems of the body which can be measured to provide biofeedback are the neuromuscular system, the respiratory system and the cardiovascular system, while biomechanical biofeedback involves measurements of movement, postural control and force. Biofeedback may also be classified according to the type of signal used, however for the purpose of this review, the classification presented in Figure 1 will be used.

Other types of biofeedback which exist, such as electroencephalography, which provides information on brain wave activity and galvanic skin response, which measures skin conductance, are outside the scope of this review and therefore will not be discussed. This review will also only discuss real-time biofeedback applications, therefore offline applications, such as blood pressure monitoring have not been addressed.

The neuromuscular system is the nervous and musculoskeletal system working together to produce movement. Any measure of these systems can be used to provide neuromuscular biofeedback. Neuromuscular biofeedback methods used in physical rehabilitation include EMG biofeedback and real time ultrasound imaging (RTUS) biofeedback.

RTUS send short pulses of ultrasound into the body and using reflections received from tissue interfaces, images of internal structures are produced [35] thus RTUS is capable of giving immediate real-time visual feedback of muscle activity by allowing the user to directly see the muscle changing shape/length on a display. A recent survey concluded that 81% of physiotherapist using RTUS used it as a biofeedback tool during rehabilitation [36]. It has been found that augmenting typical clinical instruction with visual feedback of the anterolateral abdominal wall using RTUS reduced the number of trials needed for subjects with [37] and without [38] low back pain (LBP) to perform the abdominal hollowing exercise. Conversely Teyhen and colleagues [39] reported that the addition of RTUS biofeedback did not enhance the ability of participants with LBP to perform the abdominal hollowing exercise over those who had not received biofeedback. Reports suggest that RTUS used to provide visual biofeedback improves activation of the multifidus muscle in healthy subjects [40]. RTUS has also been successfully used to provide visual feedback of pelvic floor muscle activation. Dietz et al. [41] showed that 32 of 56 women learned correct activation of their pelvic floor muscles with less than 5 minutes of RTUS biofeedback training. Ariail et al. [42] reported on the use of RTUS in the retraining of the pelvic floor muscle in a single case postpartum and concluded that the use of RTUS was a helpful biofeedback tool for re-education and rehabilitation of the pelvic floor muscles for this patient. While RTUS is widely used in physical medicine and rehabilitation, further large RCTs are required to examine its role as a biofeedback tool in physical rehabilitation.

HR can be measured using a heart rate monitor or an electrocardiogram to deliver feedback to patients. HR biofeedback is a therapeutic approach which allows patients to control their HR by means of direct representation of the numerical value of HR on a wearable device such as a watch or a handheld display. Early studies suggest that HR biofeedback could significantly lower mean HR and systolic blood pressure during treadmill exercise [43]. The results of a small study by Fredrikson and Engel [44] found that HR biofeedback resulted in a significant decrease in HR while exercising on a cycle ergometer, however systolic blood pressure was unaffected by the feedback. More recently, Moleiro and Cid [45] investigated the effects of HR biofeedback training on the control of HR during a physical exercise test, comparing it to verbal instructions to reduce HR. They found that the participants who trained with HR biofeedback showed a greater attenuation in the increase in HR produced by exercise than participants who trained with verbal control instructions. Recently HR biofeedback has been investigated as a means of controlling blood pressure in untreated hypertensives. This pilot investigation examined the effect of a short HR biofeedback protocol on the control of blood pressure and found that the systolic blood pressure and mean arterial pressure responses to an emotional speech test were significantly smaller in the biofeedback training group than in the control group who underwent blood pressure monitoring [46]. HR is a widely used measure in clinical practice, however the research that exists to support the use of HR biofeedback is limited.

HRV refers to the variability in the time between heart beat. These variations in HR are regulated by the autonomic nervous system. HRV at the frequency of respiratory, which is also termed RSA, refers to the increase in HR with inspiration and the decrease in HR with expiration [47]. HRV is easily measured and relatively reliable and thus it has been used as an index to understand a person’s internal state [48]. HRV and RSA both provide biofeedback on the cardiovascular system and both terms are used interchangeably in the literature. HRV biofeedback appears to be a useful adjunct in the treatment of asthma and may help to reduce dependence on steroid medications [49]. Preliminary data suggests that HRV biofeedback can be used to improve overall functioning and depression in patients with fibromyalgia [50]. Giardino et al. [47] examined the efficacy of an HRV biofeedback and pulse oximetry biofeedback intervention on functioning and quality of life in patients with chronic obstructive pulmonary disease. Promising findings from this report include improvements in walking and quality of life markers, however this intervention was not compared to a control therefore further investigations are necessary to draw conclusions. Research with clinically depressed individuals indicates that RSA biofeedback training facilitates an increase in HRV amplitude and a decrease in depressive symptoms [51]. Preliminary evidence also exist to support the efficacy of RSA biofeedback in improving physiological and psychological health for individuals with posttraumatic stress disorder [52]. Reports also support the efficacy of HRV biofeedback in improving symptoms and quality of life in patients with coronary heart disease [53]. A similar study by Luskin et al. [54], demonstrated that eight sessions of HRV biofeedback produced reductions in perceived stress and improved function on the 6-minute walk test in patients with heart failure. While HRV/RSA biofeedback is a relatively new area of research, preliminary observations suggest that it may be useful in improving symptoms and quality of life in a range of health conditions.

Inertial sensing uses accelerometers and gyroscopes to estimate three-dimensional (3-D) kinematic information of a body segment, such as orientation, velocity and gravitational force. An accelerometer measures acceleration and gravitational acceleration, while a gyroscope is used to measure angular velocity [62]. These inertial sensor parameters are used as input to a feedback system that delivers a wide variety of forms of feedback to the user including, auditory, visual and tactile signals. As a result of their small size and portability inertial sensors have proven useful in movement and balance applications.

A number of researchers have investigated the role of inertial based sensing biofeedback in balance training. Davis and colleagues [63] used gyroscopic measurements to provide biofeedback and found significant changes in trunk angular displacement in both young and older participants during a number of balance tasks compared to control treatment. Verhoeff et al. [64] also examined the effects of a gyroscopic biofeedback system on trunk sway during dual tasking (performing a cognitive and a motor task) while walking. Similar to Davis and Colleagues they enrolled both elderly and young participants in their study and found that the young participants were able to react to the biofeedback while walking and performing a dual task at the same time. The elderly reduced their trunk sway with biofeedback while walking normally however, when a cognitive or a motor task was added, they were less able to react to the biofeedback and reduce trunk sway. Inertial based sensing biofeedback has also been evaluated in clinical populations. Dozza and colleagues [65, 66] evaluated the effectiveness of using an audio biofeedback system based on accelerometric sensors for improving postural stability and balance in healthy subjects and in patients with bilateral vestibular loss. The first study showed that using the audio biofeedback system significantly influenced participants’ balance. The results of the study including participants with bilateral vestibular loss indicated that the audio biofeedback training reduced postural sway and was more effective for participants with bilateral vestibular loss than for the unaffected controls. A recent study by Nicolai and colleagues [67] studied whether an audio biofeedback system could be used to enhance postural control in participants with the neurological condition, progressive supranucular palsy. While improvements were demonstrated in balance, there were no significant improvements noted in measures of function. The same research group [68] more recently, investigated whether the same biofeedback system could be used to enhance postural control in participants with Parkinson’s disease. While they did observe improvements in both measures of balance and function, similar to the previous report, the results were from a small uncontrolled study with only seven participants. Soon et al. [69] examined the efficacy of using an inertial-based sensing modality as biofeedback during a tandem stance task in stroke patients. Compared with the control group, participants who received knowledge of performance biofeedback with the inertial-based sensors improved both average trunk angular displacement and velocity in trunk sway during tandem stance. While this was a small study, including only six participants per group, the improvements were shown to persist one month after the intervention.

Research has shown that sensor based feedback can be used to modify movement or behaviour. Breen et al. [70] used a biofeedback system which used a single accelerometer to correct neck posture in computer users. Crowell and colleagues [71] found that individuals can use real-time feedback of tibial acceleration from an accelerometer to reduce loading on their lower extremities while running and they can maintain the reductions for at least ten minutes after the feedback is removed.

Inertial measurement units have also been used to monitor physical activity. Much work has been conducted evaluating pedometers as a means of improving exercise compliance in people with diabetes, obesity and congestive heart failure [72‐74]. Accelerometers are particularly useful in providing objective feedback of ambulatory activity to investigators and study participants in exercise adherence research [75]. Koizumi et al. [76] evaluated the efficacy of accelerometer-based feedback on daily physical activity in community-dwelling older women and found that using accelerometers can significantly improve the quantity and quality of daily physical activity as well as cardiorespiratory endurance.

Inertial sensors have gained popularity due to their small size and portability, making them suitable for use outside the laboratory setting. While further research is required, preliminary reports have used inertial sensor biofeedback effectively to retrain balance, to modify movement and to monitor physical activity.

Electrogoniometry allows measurement of joint kinematics during functional tasks and movements yielding real-time feedback to clinicians and patients. As the kinematics of the joint change feedback is delivered, usually via an auditory signal or visual display. Ceceli et al. [88] and Morris et al. [89] analyzed the effectiveness of providing kinematic biofeedback of the knee, using electrogoniometers compared with conventional physiotherapy in efforts to minimize genu recurvatum in participants who had a CVA. Ceceli et al. [88] found that participants who were provided with the kinematic biofeedback showed a statistically significant decrease in the number of knee hyperextensions compared with those who had received conventional physiotherapy only. In the study by Morris et al. [89] participants received treatment in two separate phases; participants in the experimental group received kinematic biofeedback during the first phase and conventional physiotherapy during the second phase. Control group participants received conventional physiotherapy during both phases. This investigation showed a moderate effect for increased gait speed in the experimental group but no effect in the control group after the first phase of treatment. While both groups demonstrated statistically significant reductions in peak knee extension during stance, there was no difference noted between the groups. Colborne et al. [90] also used an electrogoniometer to provide CVA participants with kinematic biofeedback of ankle joint movement in an attempt to improve ankle control during gait. The authors found that kinematic biofeedback training resulted in a moderate increase in gait speed while conventional physiotherapy resulted in only a small improvement. However Kuiken et al. [91] examined the effects of a computerised biofeedback knee goniometer on patients’ compliance with active range of motion exercises after total knee arthroplasty and concluded that the feedback provided by the device did not have a significant influence on the rate of exercise performance. While electrogoniometery is a relatively inexpensive method of providing kinematic biofeedback, the overall benefit of using this technology has yet to be proven.

A pressure biofeedback unit (PBU) is a tool developed to aid the retraining of muscle activity and can provide useful visual biofeedback during treatment [92]. A PBU consists of an inflatable cushion which is connected to a pressure gage, which displays feedback on muscle activity. These units are relatively inexpensive and this technique is more easily applied in the clinical setting in comparison to previously mentioned techniques. PBUs have been used to indicate correct contraction of the transversus abdominis muscle during the abdominal hollowing exercise. Ciarns and colleagues [92] used a PBU in their study to quantify abdominal muscular dysfunction in participants with LBP or a history of LBP. PBUs have also been used to assess the deep cervical flexor muscles in individuals with and without neck pain [93]. Hudswell and colleagues [94] used a PBU in their study to measure deep neck flexor strength during the cranio-cervical flexion test. However no research comparing a PBU intervention to a control in the treatment of neck pain has been found. Research has found that lumbar spine stabilisation using a PBU results in significant increases in gluteus medius and internal oblique activity during a hip abduction exercise [95]. While the PBU is a useful tool for assessing the abdominal drawing in exercise, other research investigating this biofeedback tool is limited.

Video cameras allow clinicians and patients to examine locomotion qualitatively, whereas optical motion capture systems allow for quantitative 3-D movement analysis. Optical motion capture systems use a network of cameras to detect a series of markers placed on anatomical landmarks on a subject’s body. This information is then used by the system to deliver visual feedback of movement and posture.

Video camera feedback has been investigated in rehabilitation. Kim and colleagues [96] investigated the effects of using a video camera to provide visual feedback to participants with winged scapula during a push up exercise. Providing visual biofeedback resulted in increased activity of the serratus anterior muscle and decreased activity of the upper trapezius muscle. Research has also shown that using videotape biofeedback is an effective instructional method for enhancing motor skill acquisition in a post stroke population [97]. Despite the high accuracy to assess position, optical motion capture systems are generally restricted to a laboratory environment and a dearth of evidence exists to establish their role as a biofeedback tool in rehabilitation.

Biofeedback is most commonly delivered using visual, auditory or haptic signals however recent years have witnessed the emergence of immersive, VR biofeedback signals. VR and therapeutic exergames provide patients opportunities to engage in meaningful, intensive, enjoyable tasks related to real-life interests and activities of daily living [98]. A small case study reported on the use of computerized biofeedback training in a VR environment to improve hand function in a post CVA population [99]. Participants received haptic, visual and auditory feedback as they performed hand exercises, with graphics displayed on a personal computer screen. Each of the three participants studied showed improvement in measures of hand function following the two week training program. Broeren et al. [100] made use of a game in a virtual environment, along with a force-feedback haptic device, to improve control of a CVA patient’s left hemiparetic arm. Their results showed that the patient was motivated to practice and exhibited improved dexterity, grip force, and motor control. Betker et al. [101] found that centre of pressure (CoP) biofeedback controlled by video game–based exercises could improve dynamic balance control in three cases with various neurological disorders. This study also reported that the CoP-controlled video game–based exercise regime motivated subjects to increase their practice volume and attention span during training. Although these result are taken from studies with small sample sizes, the evidence suggest a role for VR based biofeedback in neurological rehabilitation. A larger study, conducted by Piron and colleagues [102] demonstrated that VR based biofeedback could enhance upper limb function in CVA participants. Subsequent to this study, Piron et al. [103] conducted a trial in which CVA participants received reinforced feedback in a virtual environment for the upper limb or a control intervention of conventional rehabilitation. Following the intervention period only the intervention group demonstrated significant improvements in measures of function. There was however no difference observed between the groups. Nevertheless this data suggests that the recovery of arm motor function in patients after recent CVA is promoted by an augmented feedback strategy, which is applied through a virtual-environment. Crosbie et al. [104] conducted a RCT to compare VR mediated upper limb therapy to standard therapy in a post CVA population. Both groups reported changes to their upper limb activity levels, however there was no difference between the groups following the intervention.

Preliminary evidence suggests that biofeedback via an exergame may be used to enhance exercise technique. Doyle et al. [105] showed that interacting with a game incorporating simple visual feedback results in improved exercise accuracy compared to performing the exercise from memory or with limited feedback in the form of an instructional video demonstration. Fitzgerald et al. [106] examined the effects of providing feedback during wobble board exercises on postural stability. Healthy adults were randomly assigned to either an exergaming group, who received visual biofeedback using a therapeutic exergaming system or to a control group, who received no feedback during exercise. While both groups in this study showed similar improvement in measures of postural stability, a greater level of interest and enjoyment was observed in the exergaming group, suggesting that there may be potential benefits to using exergaming biofeedback in the rehabilitation setting.

Recent years have witnessed huge development in the field of VR and therapeutic exergaming. Consequently VR and therapeutic exergaming biofeedback is an emerging field. The early evidence gathered was adapted from small case studies however a number of larger studies and RCTs have been conducted more recently. The existing evidence suggests a role for VR based biofeedback and exergaming biofeedback in rehabilitation, however further work is required before definite conclusions can be drawn.