Proposed step training with auditory feedback for post-stroke hemiparetic patients with knee hyperextension

Abstract

[Purpose] The aim of this study was to evaluate the efficacy of a treatment combining auditory feedback regarding knee flexion angle with gait training utilizing proprioceptive feedback in reducing knee hyperextension, focusing on knee joint angles in gait. [Participants and Methods] Six post-stroke hemiparetic patients underwent the intervention, and we measured the knee flexion angle and the shank tilt angle while they were walking with their own ankle-foot orthoses pre- and post-intervention and evaluate the therapeutic effects. [Results] The knee flexion angles in each phase of the gait cycle did not differ significantly between pre- and post-intervention, but the forward tilt angles of the lower leg in the mid-stance phase and the terminal stance phase were significantly higher post-intervention. [Conclusion] These findings suggest that the treatment described in this study alleviated knee hyperextension.

INTRODUCTION

Stroke is a global health issue, and the number of people affected by this condition is estimated to be approximately 94 million as of 2025¹⁾. Between 20% and 68% of post-stroke hemiparetic patients who are able to walk are reported to present with knee hyperextension in gait^2,3,4). Previously shown causes of knee hyperextension in gait include muscle weakness, numbness, limited ankle mobility, and proprioceptive disorders in the affected leg⁵⁾. Knee hyperextension in gait leads to decreases in gait velocity and efficiency, increases in energy consumption in gait, and deformity and associated chronic pain of the knee^{6, 7)}. In addition, pain and decreases in gait velocity are directly linked to a decline in quality of life, and thus the treatment of knee hyperextension in gait is important. Various treatments, including functional electrical stimulation of muscle on the affected side⁸⁾ and the use of orthoses⁹⁾ have been studied with the aim of treating knee hyperextension in gait. A study of 26 post-stroke hemiparetic patients who were assigned to either an 8-week motor relearning program only or an 8-week motor relearning program with auditory feedback of the knee joint angle showed a significant reduction of knee hyperextension in the group with both the motor relearning program and auditory feedback¹⁰⁾. A randomized controlled study of 32 post-stroke hemiparetic patients with knee hyperextension compared the control group who received conventional physiotherapy and the intervention group who received conventional physiotherapy with proprioceptive training involving stimulation of knee proprioception by repetitive knee flexion (e.g., walking and squatting), and found a significant 4° reduction in knee hyperextension in gait in the intervention group¹¹⁾. In addition, various feedback methods and devices have been studied. It was shown that auditory feedback was useful for learning the velocity and angles of joint motions and was effective for reducing dependence on visual input and focus on proprioception¹²⁾. Furthermore, auditory cueing was helpful for learning tempo in exercises such as walking¹³⁾. Other studies examined auditory biofeedback devices that notified the user of knee hyperextension detected by a goniometer attached to the knee^{10, 14)}. A systemic review of treatment of knee hypertension in gait recommended the integration of auditory feedback of knee flexion angle and proprioceptive training wherein patients repeatedly flex their knees at a certain angle in order to stimulate proprioceptors around the knee¹⁵⁾. However, post-stroke hemiparetic patients tend to have weak knee extensor muscles and a risk of knee buckling during knee flexion, making accurate and repeated knee flexion difficult. We therefore developed a treatment for knee hyperextension combining a knee, ankle and foot orthosis (KAFO) with an adjustable range of knee flexion, a device providing auditory feedback on knee joint angle, and gait training involving repetitive knee flexion.

The aim of this study was to evaluate the effectiveness of a treatment combining auditory feedback on knee flexion angle with proprioceptive-based gait training in reducing knee hyperextension, focusing on knee joint angles in gait.

PARTICIPANTS AND METHODS

Measurements were performed between July 2024 and August 2025. Participants were 6 hemiparetic patients in convalescent phase who were able to walk with a foot-ankle orthosis (AFO) and presented with knee hyperextension (Table 1). Inertial sensor units (AMWS020B, ATR-Promotions, Inc., Kyoto, Japan) were placed on the front of the thigh and the lower leg, and the x-axis of the device was aligned with the vertical axis. Measurements were analyzed using data analysis software (Sensor Data Analyzer, ATR-Promotions, Inc.) to calculate the knee flexion angle and shank tilt angle relative to the initial contact position in gait (0°).

Table 1. Participant characteristics.

Participant no.	Sex	Age (years)	Height (cm)	Weight (kg)	BMI (kg/m2)	Stroke type	Affected side	Days post-onset	Brunnstrom stage (lower limb)	FIM
1	Man	62	175	60	19.6	Infarction	Left	92	III	43
2	Woman	50	158	64	25.6	Hemorrhage	Right	163	III	98
3	Man	44	170	71	24.6	Hemorrhage	Left	124	V	101
4	Woman	55	153	63	26.9	Hemorrhage	Right	107	IV	116
5	Man	45	172	72	24.3	Infarction	Left	62	IV	115
6	Man	48	175	95	31.0	Hemorrhage	Right	62	IV	78
Average		50.7	167.2	70.8	25.3			101.7		91.8
SD		6.8	9.4	12.7	3.7			38.8		27.6

FIM: functional independence measure; SD: standard deviation.

This study used a personalized KAFO with adjustable knee flexion joint up to 30°, which was developed by Murayama and colleagues for safe and effective knee flexion training in post-stroke hemiparetic patients with knee hyperextension^{16, 17)}. An auditory feedback device, more precisely, a joint angle buzzer (JAB; utility model registration number 3227545) attachable to the knee joint of the KAFO, was also developed by Murayama (Figs. 1 and 2). The JAB was silent when the knee flexion angle was 0–10°, but produced a low tone sound when the knee flexion angle was 10–20° or a high tone sound when the knee flexion angle was 20–30° (Fig. 3). The device was used for step training, focusing on the knee flexion in the period between the initial contact and the loading response phase of the gait cycle, during which the normal knee joint angles were estimated to be in the range of 10–20°, and thus, the knee joint angle of 0–10° or 20–30° was considered to be hyperextension and hyperflexion, respectively. By using this device, the participants were alerted by the buzzer if they flexed the knee during step training, and they could know whether or not the knee flexion was in the appropriate range or excessive according to the buzzer tone.

Fig. 1.

Attachment of the Joint Angle Buzzer. It can be magnetically attached to the knee joint of the knee, ankle and foot orthosis.

Fig. 2.

Use of the Joint Angle Buzzer during step training.

Fig. 3.

Function of the Joint Angle Buzzer. The auditory signal changes depending on the angle range, providing feedback on knee joint angle to the participant.

Participants underwent 20-min step training, which simulated the stance phase of gait, with a KAFO and a JAB for 2 weeks with assistance of a physical therapist. We performed measurements while the participants walked at a comfortable speed for approximately 10 m with an AFO and without assistance of a physical therapist before and after the 2-week intervention. The knee joint angle and the shank tilt angle in the sagittal plane, measured and calculated using the inertial sensor units attached to the thigh and lower leg, were used as parameters for knee hyperextension during the stance phase. In normal gait, the lower leg continuously tilted forward during the stance phase, while in gait with knee hyperextension, the lower leg transiently tilted backward during the mid-stance phase as the knee extended¹⁸⁾. It has been reported that the shank tilt angle during the stance phase in gait with an AFO is useful in examining knee hyperextension¹⁹⁾. Accordingly, the present study used the shank tilt angle as an index in the examination of knee hyperextension.

Three consecutive gait cycles were extracted by defining the initial contact point from the spike-like waveform in vertical acceleration recorded by the inertial sensor placed on the lower leg. The time of each gait cycle was normalized to 100% to obtain the average cycle of three extracted gait cycles. From this average cycle, the mean knee joint angle and the mean shank tilt angle were calculated in the following phase: loading response (2–12% of gait cycle completion), mid-stance (12–31% of gait cycle completion), terminal stance (31–50% of gait cycle completion), pre-swing (50–62% of gait cycle completion), and swing (62–100% of gait cycle completion)²⁰⁾.

The Wilcoxon signed-rank test was used to compare the mean knee joint angles and mean shank tilt angles in each phase of the gait cycle between pre- and post-intervention in each participant. JSTAT (ver. 22.1 J) was used for statistical analysis with p<0.05 indicating statistical significance.

This study was approved by the Niigata University of Health and Welfare ethics committee (approval number: 19021-230414), and informed consent was obtained from all participants after explaining the details of study using written documents.

RESULTS

The knee joint angles in all phases of gait cycle did not differ significantly between pre- and post-intervention (Fig. 4 and Table 2).

Fig. 4.

Knee joint flexion angle (average of all participants).

Table 2. Knee joint flexion angle (°) in each phase of the gait cycle pre- and post-intervention.

Gait cycle	Pre-intervention (n=6)		Post-intervention (2 weeks) (n=6)
	Median	IQR	Median	IQR
Loading response (2–12%)	2.78	5.07	2.41	5.11
Mid-stance (12–31%)	−5.77	5.36	−3.21	5.41
Terminal stance (31–50%)	−13.47	8.35	−7.15	15.34
Pre-swing (50–62%)	−3.37	16.93	5.57	12.79
Swing (62–100%)	6.04	6.84	7.92	6.96

IQR: interquartile range.

The forward tilt angles of the lower leg in the mid-stance phase and the terminal stance phase were significantly higher post-intervention compared with pre-intervention (Fig. 5 and Table 3, p<0.05).

Fig. 5.

Shank tilt angle (average of all participants).

Table 3. Shank tilt angle (°, forward positive) in each phase of the gait cycle pre- and post-intervention.

Gait cycle	Pre-intervention (n=6)		Post-intervention (2 weeks) (n=6)
	Median	IQR	Median	IQR
Loading response (2–12%)	6.60	1.51	6.58	1.96
Mid-stance (12–31%)	10.37	3.47	14.62	3.30*
Terminal stance (31–50%)	18.46	6.27	22.89	3.04*
Pre-swing (50–62%)	30.38	25.44	34.98	19.40
Swing (62–100%)	19.89	5.23	17.55	5.36

*p<0.05 (Wilcoxon test). IQR: interquartile range.

DISCUSSION

This study proposed step training with a KAFO to enable users to flex their knee safely with auditory feedback and examined its effectiveness for the treatment of knee hyperextension.

We found no statistically significant changes in the knee joint angle post-intervention, indicating that the knee joint angle alone did not detect alleviation of knee hyperextension. However, the forward tilt angles of the lower leg in the mid-stance phase and the terminal stance phase were significantly increased after intervention. The transient backward tilt of the lower leg at 20% of gait cycle completion observed pre-intervention was absent post-intervention, suggesting the alleviation of knee hyperextension.

The decreases in the backward tilt angle can be explained by the fact that participants were able to repeat knee flexion accurately using an auditory feedback system and a KAFO that allowed up to 30° knee flexion. Repetitive knee flexion training can be challenging for post-stroke hemiparetic patients with knee extensor muscle weakness and knee instability because they have a risk of knee buckling and may be afraid to move into a flexed knee position. However, in this study, the participants seemed to feel comfortable enough to move into the knee flex position, knowing that the knee joint of the KAFO would not flex more than 30°, even if their knee buckled.

Previous studies have shown that accurate repetition of rehabilitation tasks is crucial for rebuilding neural networks and motor learning after stroke^{21, 22)}. The present study used the JAB for auditory feedback, which accurately informed the participants and physical therapists about the angle of the knee joint. This enabled the physical therapists to assist the participants to repeatedly and accurately move into a flexed knee position, making the step training effective even during a short period of just 2 weeks. The alleviation of knee hyperextension shown in this study may address issues associated with knee hyperextension, including decreases in gait velocity and onset of pain, as pointed out in previous studies^{6, 7)}, and may also lead to an improvement in quality of life in post-stroke hemiparetic patients.

This study has some limitations that should be noted. First, the direct reasons for the observed decreases in the backward tilt angle of the lower leg were not examined. Detailed studies involving electromyography, three-dimensional motion analysis, and other methods are necessary to investigate knee extension moment and changes in how knee extensors (e.g., quadriceps femoris) function. Other limitations include the limited number of participants, the lack of a control group, the possibility of spontaneous recovery in hemiparetic patients in the convalescent phase, and the lack of gait velocity measurements.

After the step training with auditory feedback developed in this study, there were no significant differences in knee joint flexion angle in each phase of gait cycle between pre- and post-intervention, but we found significant increases in the forward tilt angle of the lower leg in the mid-stance and terminal stance phases, suggesting that this intervention alleviated knee hyperextension.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

The authors declare no conflicts of interest.