Worldwide, stroke is ranked as one of the leading cause of disability [1]. Up to 88% of individuals suffering stroke experience hemiparesis with disorders of gait and balance; which may persist even in the chronic phase [2, 3]. Therefore, a crucial goal in the rehabilitation of patients affected by stroke is to restore their mobility in order to allow resumption of independent living and improved quality of life [4, 5]. Post-stroke gait suffer from reduced abilities due to balance, speed, symmetry and muscle strength deficiencies [6]. Individuals suffering sub-acute and chronic stroke continue to use assistive or supportive devices in order to overcome postural control deficiencies, gait asymmetry, sensorimotor deficiencies and speed deterioration to increase their gait ability [7, 8]. Recent studies show that rehabilitation and gait training may also benefit chronic hemiplegic patients [9–11]. Involvement of a physical therapist is necessary to administer gait rehabilitation procedures, but due to the high incidence of stroke, the patient to therapist ratio is increasing and creating a situation which demands the increased use of engineering technology to aid therapists in gait rehabilitation interventions [12]. Gait rehabilitation after stroke requires intensive task-related training as well as variable training in changing environmental contexts with increasing physical demand [13]. Overground gait training with the use of assistive devices and in combination with other therapies or exercise regimens may benefit the gait rehabilitation after stroke [14].

Among the currently available gait rehabilitation systems, several recently developed devices offer gait training with overground walking for individuals suffering stroke [15–20]. These devices often use powered actuators for lower extremity joints, and may increase the post-stroke walking functionality towards a normal gait pattern. Meanwhile, the numerous limitations of the currently available systems such as high cost, wearability, weight, safety, and other issues often deny many individuals from undertaking post-stroke gait rehabilitation regimens with such devices. Therefore, it has been a difficult task to develop cost-efficient and adequate overground gait rehabilitation devices with a simple setup process and intuitive interactions targeted towards patients who can stand and move, but still need to overcome gait deficiencies.

Rehabilitation devices may also deliver sensory cues to the patients through visual, auditory and haptic modalities. These cues can be applied individually or in combination for task-oriented neuromotor rehabilitation following trauma [21]. Applications involving visual cues require a bulky setup (display monitor/surface). Therefore they are mostly not feasible for use with portable/wearable gait training devices. Auditory cues can be delivered using compact systems but are quite cumbersome for the users in gait training application, as they continuously occupy the hearing sensation and may obstruct activities of daily life. Moreover, noisy or inappropriately lit environments may also restrict the practical usability of auditory or visual cues [22]. The term haptic cue refers to the provision of information/experience to the users through their sense of touch. Visual and auditory senses are mostly occupied by receiving information to maintain the activities of daily life, whereas the haptic sensation is typically underutilized [23]. Thus, haptic cues may be delivered to the users without burdening their frequently used senses of vision and hearing. The ectoderm present at the early embryo stage is the basis of formation of both the skin and the nervous system [24]. This connection forms a link whereby the brain activity can be invoked through haptic stimuli, implying that the use of the haptic channel may provide an ease in changing neural plasticity. Haptic cues are generally divided into two categories; Tactile and Kinesthetic. Kinesthetic cues generally include a sensation of force at the location of the interface and offer a spatial frame of reference to the user. PHANTOM [25] and SPIDAR-G [26] are typical examples of devices that provide kinesthetic cues. Tactile Cues generally include the sensation of vibration, texture or pressure. These can be provided by devices such as LinkTouch [27] and MIMIC [28]. Thus, provision of haptic cues can be achieved through devices that deliver kinesthetic and tactile sensations to the user [29].

Provision of discrete kinesthetic cues to individuals with subacute stroke, through an instrumented cane, during overground walking has been found to improve paretic muscle activity [30]. Light touch (LT) refers to fingertip contact with another physical object [31]. When lightly touching a surface, proprioceptive receptors send information of position and velocity of the body sway to the cortical areas that control posture, leading to the activation of postural muscles to attenuate sway [32]. Continuous kinesthetic cues provided to individuals with chronic stroke through an instrumented haptic bar coupled with a visual display and with manipulation of physical environments improved their gait stability in challenging conditions such as downslope walking [33]. Tactile cues (muscle stimulation) provided to individuals suffering chronic stroke improved their gait performance during overground walking [34]. In addition to general physical therapy, tactile cues applied for therapy may improve gait performance in patients with chronic stroke and foot drop [35]. Thus, whether provided as kinesthetic or tactile, haptic cues may facilitate recovery of post-stroke gait during rehabilitation process. In this context, haptic cues can be a useful therapeutic option for intuitive interactions in overground gait rehabilitation; the applications of which in post-stroke ambulation recovery can be further explored.

We recently proposed a haptic cane device (HCD) which can deliver kinesthetic cue [36], and a vibrotactile feedback device (VFD) which can provide tactile cue [37]. Both HCD and VFD are low-cost and easy-to-use systems that can provide overground gait rehabilitation experience for individuals with sub-acute/chronic stroke. In the pilot study with HCD [36], we observed that proprioceptive augmentation delivered to individuals with sub-acute stroke could improve their paretic muscle activation and allow them to walk with increased gait speed on the ground while maintaining balance. Walking at increased gait speeds improves the kinematics and muscle activation patterns of hemiparetic gait [38], which is consistent with the speed-dependent changes reported in healthy subjects [39]. Increased gait speed also influences temporal symmetry of hemiparetic gait [40]. Moreover, post-stroke lower limb muscle functionality is correlated with comfortable and faster gait [41, 42]. Additionally in the pilot study with VFD [37], we observed that delivering tactile cue induces the increase in stance time of paretic lower limb, which effectively helps in reduction of temporal gait asymmetry. In studies related to upper limb, complementary integration of kinesthetic and tactile cues can improve hand movement perception [43], activate heteromodal areas to subserve multisensory integrative mechanisms at cortical and subcortical levels [44] and facilitate motor learning [45]. Novel rehabilitative approaches that combine simultaneous motor and sensory stimulations may substantially improve muscular strength and joint position sense in chronic stroke patients [46]; thereby confirming the strong impact of somatosensory stimulation on motor recovery [47]. The integration of kinesthetic and tactile cues for application to post-stroke gait rehabilitation has previously undergone little to no exploration [48]. Proprioceptive and tactile afferents both terminate and share overlapping networks in the somatosensory cortex [49]. Therefore, exploration of the effects of such cues on lower limb movements is warranted.

In this paper, we have explored the effects of combining the use of portable HCD and wearable VFD for a novel integrated cue system at various gait speeds. Our proposed system provides voluntary overground gait training modes for hemiparetic post-stroke ambulatory individuals. The first aim of this study is to determine whether the individual benefits of HCD and VFD can be observed as combined in the integrated cue system at user-preferred gait speed. The second aim of this study is to evaluate the effects of haptic cues (kinesthetic vs integrated) and increased gait speed on the characteristics of balance, gait symmetry and lower limb muscle activity.

Integrated cue system with its components and application is shown in Fig. 1. The main idea of this system is to provide over-ground walking with aid of HCD for kinesthetic perception and a vibrotactor array attached on the paretic leg to cue the gait modification through VFD. Thus, integrated cue system has two basic functions, which are velocity control of the cane and vibration on the shank during swing phase of gait. The designs of HCD [36] and VFD [37] have individually been evaluated through pilot studies, the results of which have already been published.

HCD offers a new design, composed of a motorized wheel and a cane structure to provide continuous proprioceptive augmentation [36]. Canes and walkers have traditionally been used to improve the patient’s ability of walk by increasing the base of support area and also maintaining the voluntary gait by providing a psychological sense of stability [50]. However, these traditional devices require excessive use of the upper extremities to keep balance and manage weight distribution, which causes an aggravating gait imbalance due to the reduced involvement of the lower extremities. HCD prevents the patient to use excessive support force of an upper limb as a conventional cane and encourages the patient to use paretic lower limb more actively due to the reduced weight-bearing. The structure of HCD, which consists of a wheeled base and a cane, induces the balance stability by providing a continuous proprioceptive input and increasing user’s base of support. The HCD is velocity-controlled and can therefore help regulate the user’s gait speed. This makes it a valuable tool of gait rehabilitation in post-stroke ambulatory subjects. Use of the healthy hand is recommended for hemiparetic stroke subjects to operate and walk with HCD. As the users cannot apply excessive force for the weight support on the cane with a wheeled structure, they need to use their affected lower limb instead of chiefly relying on the upper limb. This operation mode of HCD provides a user with device-driven walking over the ground similar to treadmill-driven walking on a treadmill [51].

VFD is a wearable device which provides tactile cues to the user during overground walking and serves as a tool for diagnosing temporal asymmetry of gait. An elastic belt containing an array of six vibrotactors, securely attached to the inside of the belt, is used for the tactile cue application to the subject [37]. The vibrotactors cover the entire proximal end of paretic shank from front to back. Vibration with constant intensity at 200 Hz is provided during the paretic swing phase to maximally stimulate high-frequency Pacinian mechanoreceptors [52]. Each vibrotactor operates at 3.3 V and 66 mA, producing vibration amplitude of 1.4 G. The belt is worn on the lower leg, so that it does not interfere with proprioceptor information from the ground, as reduced feedback due to proprioceptive loss is likely to impair balance [53]. Also, direct muscle stimulation may contribute in enhancing the gait modification through afferent signal of vibration [54, 55]. The on/off status of the vibration signal can be visualized using an LED connected to the array of vibrotactors. The whole system can be worn easily using elastic belts with Velcro fasteners. A pair of insoles containing four force-sensitive resistors each, positioned at the heel, toe, fifth metatarsal, and first metatarsal are incorporated in VFD system, for collection of ground contact data.

Data analysis

During each walking trial, various parameters were recorded for post-experimental data analysis. Figure 2 illustrates the communication interface of the various devices used in this research. HCD was operated over Wi-Fi with a personal computer running a custom-built LabVIEW GUI program. This GUI commanded the HCD to setup various walking speeds during trials. It also handled the smartphone communication over Wi-Fi and saved the body sway data for post-experimental analysis. Later we calculated RMS of Mediolateral (ML) tilt to determine the condition of balance during various walking trials. The tilt in Mediolateral plane (ML tilt) is the sideways sway a human body experiences while maintaining an upright stance or walking. The RMS of ML Tilt has been utilized as a reliable marker of postural control during walking [62, 63]. Stance times received from the VFD were stored on the personal computer and utilized to define Stance Symmetry Ratio (i.e., Stance Symmetry Ratio (SSR) = stance time of healthy side / stance time of paretic side). The use of wireless EMG Sensors minimized the noise and artifacts of electrical and mechanical origins. The recorded EMG signals were band pass filtered (20–400 Hz), rectified, and smoothed with RMS filter at 100 ms [64]. To identify the increase in muscle activity of paretic side percentage of non-paretic peak activity (%NPA) was calculated, where integrals of EMG values of the paretic muscles during stance and swing phase were normalized to the EMG integrals from peak activity of the same muscles on the non-paretic side [30, 65, 66].

A one-way repeated measures analysis of variance (ANOVA) was performed to study the effects of Cue (factor), at preferred gait speed, which had four levels (no cue (NCW), tactile cue (TCW), kinesthetic cue (KCW + 0%) and integrated cue (ICW + 0%)) on RMS of ML Tilt, Stance Symmetry Ratio and %NPA in the four paretic muscles. Moreover, a two-way repeated measures ANOVA was used to identify the effects of Haptic Cue (factor, levels: kinesthetic (KCW) and integrated (ICW)) and increase in Gait Speed (factor, levels: 0%, 20% and 40%) on RMS of ML Tilt, Stance Symmetry Ratio and %NPA in the four paretic muscles. In addition, Mauchly’s test of Sphericity was used and Greenhouse-Geisser corrections were applied in case of its violation. Post hoc tests were conducted using the Bonferroni correction method. Partial eta squared (η²) was calculated as a measure of the effect size for one- and two-way repeated measure analysis of variance; Cohen’s d was also calculated and used for pairwise comparisons. All statistical analysis was performed using SPSS V20.0 (IBM Corp., Armonk, NY, USA).

Our proposed system provides the user with an experience of over-ground walking. The overloading of upper limb is restricted by the design of haptic cane device (HCD) which increases involvement of lower limb paretic muscles. Contact of the user with the handle of HCD provides a continuous proprioceptive input; it serves the purpose of assisting balance. Meanwhile, the tactile cue provided by vibrotactile feedback device (VFD) enhances the gait modification through afferent signals of vibration. In this study we combined the promising prospects of kinesthetic and tactile haptic cues to comprise a novel integrated cue system.

The purpose of this paper was to identify the benefits of combining kinesthetic and tactile cues during overground walking, with preferred and increased gait speeds, delivered to individuals with unilateral hemiparetic stroke. The subjects could utilize the proposed integrated cue system at increased gait speeds, but the most significant benefits of the system were exhibited during user-preferred gait speed trials. At preferred gait speed, subjects tend to walk with increased temporal stance symmetry and paretic muscle activity without disruption in balance; thus indicating that combining the haptic cues was beneficial. Similar improvements were observed with 20% increase in the gait speed, where the improved muscle activity and gait symmetry more likely to have been induced by the increase in gait speed. On the other hand, those improvements faded away with further increase in gait speed due to the limitations of hemiparetic gait. This demonstrates that the efficacy of the proposed system is influenced by gait speed. In light of the above-mentioned observations, the integration of haptic cues may benefit post-stroke gait rehabilitation by inducing simultaneous improvement in gait symmetry and muscle activity.