Comparison of kinetic adaptations in gait initiation following exergaming and balance training in athletes with chronic ankle instability

Abstract

Gait initiation (GI) requires coordinated postural control and is often impaired in athletes with chronic ankle instability (CAI). Exergaming has been proposed as an innovative rehabilitation strategy, but its effects on GI kinetics compared with conventional balance training remain unclear. Thirty-four athletes with CAI were assigned to either an exergaming intervention or a balance training control group. Both groups underwent a 12-session intervention, with sessions held three times per week and each session lasting 60 minutes. Kinetic measures, including center of pressure (COP) displacement and velocity in both anterior–posterior (AP) and medio-lateral (ML) directions, were assessed across the anticipatory, weight transition, and locomotor phases of GI at pre-intervention, post-intervention, and one-month follow-up. During the anticipatory phase, the exergaming group demonstrated significantly higher AP COP velocities post-intervention compared with the control group (p = 0.001, η² = 0.257), although this effect was not maintained at follow-up. In the weight transition phase, both groups exhibited significant forward displacement of COP from pre- to post-intervention (p = 0.018, η² = 0.024), with partial retention at follow-up. In the locomotor phase, ML COP displacement toward the step-initiating limb decreased significantly from pre- to post-intervention in both groups (p = 0.022, η² = 0.024) and remained stable at one-month follow-up (p = 0.009, η² = 0.028). All other kinetic parameters showed no significant changes. The pooled effect size across GI variables was small but significant (Weighted Mean dppc2 = 0.20, SE = 0.10, 95% CI [0.01, 0.40]), indicating modest but measurable benefits of exergaming on GI performance. Exergaming appears to produce improvements in anticipatory postural adjustments during GI in athletes with CAI, while effects on other phases of GI were comparable to balance training. These findings suggest that exergaming may be a useful rehabilitation tool for enhancing certain aspects of postural control. However, the retention of anticipatory gains over time was limited, indicating that further research is needed to confirm the durability and generalizability of these effects.

Introduction

Ankle sprains, particularly lateral ankle sprains (LAS), are among the most prevalent musculoskeletal injuries, accounting for a substantial portion of sports-related injuries. Up to 40% of individuals experiencing LAS may develop chronic ankle instability (CAI), although reported prevalence varies depending on diagnostic criteria, population characteristics, and follow-up duration¹. CAI is commonly defined in consensus statements, including those of the International Ankle Consortium, as recurrent sprains, persistent pain, and repeated sensations of the ankle “giving way”². These symptoms negatively affect daily functioning and athletic performance and are associated with prolonged rehabilitation, reduced physical activity, and an increased risk of long-term complications such as osteoarthritis^3,4. In addition, CAI is associated with reduced quality of life and heightened fear of re-injury, which further restrict participation in physical activity and sports⁵. These functional and psychosocial consequences are largely attributed to impaired postural control, which is considered a key underlying mechanism in CAI⁶. Individuals with CAI often exhibit deficits in both anticipatory postural adjustments (APAs) and compensatory postural adjustments (CPAs)⁷. APAs are proactive, feed-forward mechanisms that stabilize posture prior to movement initiation, whereas CPAs are reactive, feedback-driven responses that restore balance after destabilization^8,9. Deficits in these mechanisms increase susceptibility to recurrent injury and highlight the need for rehabilitation strategies that specifically target postural control.

Conventional rehabilitation for CAI typically involves strengthening, proprioceptive, and balance-focused exercises designed to enhance neuromuscular control and postural stability^10,11. Although balance training is a cornerstone of CAI management, traditional exercises are often repetitive, provide limited real-time feedback, and may reduce motivation and adherence over time^12–14. Exergaming has therefore emerged as a promising alternative by integrating physical exercise with interactive gaming technology to create an engaging rehabilitation environment. Continuous visual and auditory feedback allows real-time correction of movement patterns and facilitates motor learning and postural control¹⁵. In individuals with CAI, visually challenging conditions—such as stroboscopic stimulation—have been shown to enhance dynamic balance and reduce subjective instability¹⁶. These conditions also increase neuromuscular control and brain activity by placing greater demands on visual processing and movement coordination^17,18. Incorporating visual perturbations into exergaming may therefore provide additional benefits for balance rehabilitation. Exergaming also engages cognitive and motor systems simultaneously, creating a dual-task environment that may address information-processing deficits commonly observed in CAI^13,19–23. Although this approach imposes a relatively high cognitive load during training, repeated practice can promote motor skill acquisition and more efficient sensorimotor integration, potentially reducing the cognitive effort required for postural control during functional tasks²⁴. Furthermore, exergaming involves multidirectional weight shifts at varying speeds and amplitudes, replicating coordinated ankle, hip, and trunk strategies that are essential for functional stability in sports and daily activities²⁵. Together, these features suggest that exergaming may provide advantages over traditional balance training for improving postural control in individuals with CAI.

Despite these advantages, relatively few studies have examined exergaming in CAI populations^8,26–32. Most existing studies have focused on relatively simple standing tasks, such as maintaining the center of gravity (COG) on stable or mildly perturbing surface (e.g., Biodex platform)^30,31, or have relied on clinical balance assessments such as the Star Excursion Balance Test (SEBT), Y Balance Test (YBT), or Single-Leg Stance Test^8,26–29. More recent investigations have incorporated more complex dynamic tasks, including ball-kicking and jump-landing, which impose greater demands on neuromuscular and postural control systems^8,32. While exergaming has generally been associated with improvements in static and dynamic balance, APAs, and task-specific biomechanical control, comparisons with traditional rehabilitation approaches remain inconsistent. Some studies report superior outcomes following exergaming^26–31, whereas others observe no significant between-group differences^8,32, with traditional balance training occasionally demonstrating advantages in muscle activation and postural strategies⁸. These inconsistencies appear to be strongly influenced by methodological differences, particularly the complexity of the assessment tasks. Exergaming tends to show clearer benefits during simpler tasks, whereas advantages are less pronounced during more complex dynamic conditions, underscoring the importance of selecting functionally relevant and sufficiently challenging outcome measures.

Gait initiation (GI) is a functional relevant task that places substantial demands on dynamic postural control. In athletes, efficient GI is essential for safely transitioning from a double-leg to a single-leg stance and underlies more complex sport-specific movements such as rapid directional changes, jumping, and landing. GI occurs frequently in both daily life and athletic activities and requires coordinated forward progression while transitioning from a stable to a less stable posture³³. GI involves a well-defined anticipatory postural phase, during which APAs shift the center of pressure (COP) backward and toward the stepping limb to facilitate forward and lateral displacement of the center of mass (COM), followed by a locomotor phase that generates forward momentum^34,35. In individuals with CAI, altered COP trajectories during GI—characterized by reduced displacement and velocity—have been reported, indicating impaired APAs and inefficient dynamic balance control^36–39. Consequently, interventions that enhance COP displacement and velocity during GI may provide clinically meaningful improvements in dynamic postural control.

Despite growing interest in exergaming for CAI rehabilitation, the underlying biomechanical mechanisms through which it may improve dynamic postural control remain unclear. In particular, no study has yet investigated its effects on GI biomechanics and APAs, which are critical and functionally relevant components of postural control in athletic populations. Accordingly, this study aimed to evaluate whether exergaming provides additional benefits over conventional balance training in improving GI biomechanics and anticipatory postural control. We hypothesized that exergaming, through interactive feedback, cognitive–motor engagement, and multidirectional functional movements, would lead to greater improvements in APAs and GI parameters compared with traditional balance training.

Method

Study design

This study was designed as a randomized, single-blinded, controlled trial. Participants were allocated to either an exergaming group or a balance training group. Outcome measures were evaluated at three time points: pre-intervention, post-intervention, and one month after the intervention. Post-intervention assessments were conducted within 24 to 48 hours of completing the treatment. Dependent variables consisted of COP-related parameters in the anterior–posterior (AP) and medial–lateral (ML) directions, including displacement and velocity, measured across the phases of GI (anticipatory, weight transition, and locomotor). All sessions were delivered individually in a controlled laboratory setting. The study was registered as a clinical trial in the Iranian Registry of Clinical Trials on June 29, 2024 (IRCT20230124057209N2).

Participants

Athletes with CAI were recruited from sports physiotherapy clinics through posters and social media advertisements (Table 1). A CONSORT-based flow diagram depicts the progression of participants from enrollment through to analysis (Figure 1). All participants provided written informed consent, and the study was approved by the Ethics Committee of Tehran University of Medical Sciences (IR.TUMS.FNM.REC.1401.146). All methods were performed in accordance with the relevant guidelines and regulations.

Table 1.

The mean ± standard deviation (SD) of demographic and clinical characteristics.

Characteristics	Exergaming (N=17)		Control (N=17)		P
	Mean	SD	Mean	SD
Age (years)*	22.71	3.69	27.53	8.29	0.114
Height (cm)	176.53	7.24	174.00	9.61	0.393
Weight (kg)	69.59	9.79	70.18	11.18	0.871
CAIT (0–30)	16.28	4.34	15.33	5.30	0.570
Time since first sprain (months)	16.12	3.59	16.71	4.04	0.657
Time since last sprain (months)*	4.94	3.47	6.76	4.15	0.306
Number of giving ways	8.18	4.69	6.94	4.16	0.423
Physical activity (hours/week)	10.79	4.98	9.00	5.02	0.303
Female (N (%))	3 (18)		8 (47)		0.067
Right affected side (N (%))	9 (53)		9 (53)		1.000
Right dominant side (N (%))	14 (82)		15 (88)		0.628

CAIT = Cumberland Ankle Instability Tool.

 Chi-square test was used.

 Independent t-test was used.

^*Mann-Whitney U test was applied.

Fig. 1.

CONSORT flowchart.

Participants were eligible if they met criteria for CAI aligned with the recommendations of the International Ankle Consortium, including age 18–40 years; self-reported history of an initial acute ankle sprain occurring more than 12 months prior to enrollment; most recent sprain at least 3 months before enrollment; at least two episodes of “giving way”, recurrent sprains, or sensations of ankle instability within the past 6 months; and a Cumberland Ankle Instability Tool (CAIT) score ≤ 24 (lower scores indicate greater instability)². Participants were required to be recreationally active, engaging in sports such as volleyball, basketball, soccer, or handball at least three times per week for a minimum of 30 minutes per session⁴⁰. Individuals with a history of lower extremity fractures or surgery, known psychological or neurological disorders, or prior experience with video game–based training were excluded. Participants were also excluded if they missed three or more non-consecutive or two consecutive training sessions, or were unable to perform the required exercise maneuvers.

Sample size

As no prior studies had been conducted in individuals with CAI, the sample size was determined based on a previous study that examined the effects of exergaming on GI in older adults²⁵. An a priori power analysis was performed using G*Power 3.1.9.2 to achieve 80% power. A repeated-measures ANOVA with a within–between interaction design was specified. The calculation was based on effect sizes of 0.06 and 0.19 for ML COP displacement and velocity at anticipatory phase of GI, respectively, with α = 0.05, power = 0.80, a non-sphericity correction (ε) of 1, and a correlation among repeated measures of 0.5. The results indicated that a minimum sample of 28 participants was required to detect a within–between interaction effect in a two-group, three-measurement design. To accommodate a potential 20% dropout, 34 participants were recruited.

Randomization and blinding

Participants were randomly assigned (1:1) to the intervention (exergaming) or control (balance training) group using block randomization (block size = 4) via a web-based service. An independent third party conducted the randomization, and group assignments were concealed in sequentially numbered, opaque envelopes. Outcome assessors were blinded to group allocation.

Procedure and data acquisition

Participants stood barefoot in a relaxed posture on a force platform (Bertec Corporation, Columbus, OH, USA), where ground reaction forces (GRFs) and COP trajectories were recorded. They maintained their eyes open with body weight evenly distributed and were instructed to initiate forward stepping with the affected limb in response to a go stimulus³⁵. To ensure natural movement patterns, no restrictions were placed on step length or velocity. Each GI task was repeated five times to improve reliability, and the average value across the five trials was analyzed⁴¹.

For data processing, raw signals were low-pass filtered using a 6th-order Butterworth filter with zero-phase shift and a cut-off frequency of 10 Hz. GI phases were determined based on three gait events: onset, Landmark 1 (L1), and Landmark 2 (L2). The onset of COP displacement was defined as the point at which the vertical GRF exceeded the mean + 2 SD of the first 500 ms of quiet stance^34,35. L1 was defined as the most posterolateral COP position beneath the step-initiating limb, whereas L2 corresponded to the most lateral COP position beneath the contralateral stance limb. Based on these events, three distinct phases were extracted¹: The anticipatory phase, from COP onset to L1; (2) the weight transition phase, from L1 to L2; and (3) the locomotor phase, from L2 to toe-off of the stance limb (Figure 2)^34,35.

Fig. 2.

Schematic representation of the center of pressure (COP) trajectory during gait initiation. The onset of COP displacement, landmark 1 (L1), and landmark 2 (L2) were identified, dividing the COP trajectory into three phases: Phase 1 (anticipatory), Phase 2 (weight transition), and Phase 3 (locomotor). The swing limb refers to the limb initiating gait.

The displacement and velocity of the COP in the AP and ML directions were then computed using the following equations, where n represents the number of data points, Fs is the sampling frequency (500 Hz), and X and Y denote COP positions in the ML and AP directions, respectively^34,35.

Although electromyography (EMG) provides a direct measure of muscle activation during APAs, the present study assessed APAs using COP metrics, as the primary focus was on whole-body postural control during GI. APAs can be evaluated with various tools, including EMG, Kinetic (COP), and kinematic measures, all of which are suitable because they capture preparatory postural events occurring prior to movement onset⁴². COP-derived parameters, such as displacement and velocity during GI, have been widely applied to quantify APAs in diverse populations, including older adults⁴³, individuals with Parkinson’s disease⁴⁴, stroke patients⁴⁵, and those with chronic ankle instability³⁹. These measures provide meaningful insight into anticipatory postural control and its underlying neuromuscular strategies.

Groups

The study included two groups: an exergaming group and a balance training group. Both groups underwent a 12-session intervention, with sessions held three times per week and each session lasting 60 minutes.

Exergaming

Exergaming was delivered using a Wii Balance Board (Nintendo Co. Ltd., Kyoto, Japan) and included games such as Single Leg Extension, Torso Twist, Single Leg Twist, Sideways Leg Lift, Rowing Squat, Table Tilt, Penguin Fishing, Soccer Heading, Tightrope Walk, and Snowboard Slalom. Game order in each session was determined according to participants’ preferences, with each game lasting six minutes to allow all games to be completed per session, as previously described in related exergaming research^8,32. Each game offered three difficulty levels (beginner, advanced, expert), each with four sub-levels (unstable, amateur, professional, champion), and participants advanced to higher levels upon successful completion. Progression was automatically managed by Wii Fit, with illuminated stars indicating success, and participants earned additional stars and points for superior balance performance.

Some games required participants to maintain their COG within a designated area, while others involved guiding a game character to a target by shifting weight in multiple directions at controlled speeds and magnitudes. These tasks imposed significant cognitive demands, necessitating focused attention and rapid responses to visual stimuli. Specific games, such as Table Tilt and Penguin Slide, required planning, whereas Soccer Heading involved decision-making and response inhibition^23,46. Collectively, the selected games integrated both cognitive and motor components.

Within Wii Fit, balance-focused games such as Single Leg Extension, Torso Twist, Single Leg Twist, Sideways Leg Lift, and Rowing Squat required participants to maintain postural stability while performing dynamic movements, thereby challenging their ability to control balance during motion^8,32. This aspect is particularly relevant for athletes and individuals recovering from ankle injuries, as it reflects movement patterns and challenges encountered in sports and daily activities. Additional games, such as Snowboard Slalom, Table Tilt, Penguin Fishing, Soccer Heading, and Tightrope Walk, incorporated multidirectional weight-shifting tasks that closely replicate real-world movements, thereby enhancing dynamic stability and confidence. Several of the selected games, including Rowing Squat, required simultaneous activation of multiple muscle groups, promoting muscular strength, coordination, and endurance. This integrated approach supports rehabilitation by targeting postural control, muscular strength, and motor coordination concurrently, effectively preparing participants for the physical demands of athletic performance and daily activities^8,32. Detailed information about the games is provided in Table 2.

Table 2.

The description of games.

Games	Explains
Single leg extension	The participant balanced on one leg while swinging the opposite leg forward and backward and moving the upper limbs in opposite directions, ensuring that the COG stayed within a designated yellow circle shown on the screen.
Torso twist	The participant sustained balance while twisting their upper body, ensuring that the COG remained within a specified yellow circle displayed on the screen.
Single leg twist	The participant sustained balance while lowering one hand down and raising the opposite knee up to touch it with the back of the hand, ensuring that the COG remained within a specified yellow circle shown on the screen.
Sideways leg lift	The participant maintained balance on one leg while lifting the other leg sideways and raising the opposite arm, ensuring that the COG remained within a designated yellow circle shown on the screen.
Rowing squat	The participant balanced on their legs, pushing their hips back and bending their knees to about 120 degrees while simultaneously pulling their arms in towards their sides. The goal was to synchronize the squatting motion with the rowing movements. Visual cues displayed on the screen assisted players in maintaining proper form and timing, and they received feedback on their performance. This exercise targets various muscle groups, including the legs, core, and upper body, to improve strength, endurance, and coordination.
Table tilt	The participant shifted their weight forward and side to side to align their COG with the base of support, guiding balls into corresponding holes.
Penguin fishing	The participant shifted their weight forward and sideways to adjust the COG over their base of support, allowing them to control the movement of a penguin on ice and catch a fish.
Soccer heading	The participant adjusted their COG to head the virtual soccer ball while striving to avoid colliding with other objects.
Tightrope walk	The participant maintained rhythmic weight shifting to walk steadily without stumbling, while also navigating around obstacles.
Snowboard slalom	The participant shifted their weight from side to side to guide the snowboarder through a series of gates as quickly as possible while maintaining balance and avoiding obstacles.

COG = Center of Gravity.

Balance training

The balance training program included exercises such as single-limb stance, single-limb stance with ball kicking, single-limb hop to stabilization, and hop to stabilization with reach. Further details of the training protocol are provided in Table 3^{8,32,35,47,48}. Participants progressed to higher levels only after successfully completing the current level without errors, with the specific number of error-free repetitions required outlined in Table 3. Throughout the sessions, the therapist provided detailed verbal guidance to identify mistakes and reinforce correct performance, gradually reducing instructions over time to promote independent and accurate execution of the exercises.

Table 3.

The description of balance training.

	Difficulty Levels	Instructions	Criteria for Progression	Errors
Single-Limb Stance	1) Eyes open, hard surface, 30 s, 3 Reps 2) Eyes open, hard surface, 60 s, 3 Reps 3) Eyes open, foam surface, 30 s, 3 Reps 4) Eyes open, foam surface, 60 s, 3 Reps 5) Eyes open, foam surface, 90 s, 3 Reps 6) Eyes open, foam surface, 30 s, ball toss, 3 Reps 7) Eyes open, foam surface, 60 s, ball toss, 3 Reps 8) Eyes open, foam surface, 90 s, ball toss, 3 Reps 9) Eyes closed, hard surface, 30 s, arms out, 3 Reps 10) Eyes closed, hard surface, 30 s, arms across, 3 Reps 11) Eyes closed, foam surface, 30 s, arms out, 3 Reps 12) Eyes closed, foam surface, 30 s, arms across, 3 Reps	The participant attempted to maintain balance on the affected limb while facing challenges such as altering the base of support, closing eyes, extending the duration, and throwing a ball.	If the participant could complete 3 repetitions without errors at each level of difficulty.	Errors were defined as: A. Touching the ground with the opposite foot. B. Excessive trunk motion (more than 30 degrees of lateral flexion). C. Resting the opposite limb against the stance limb. D. Lifting the hands from the chest while standing.
Limb Stance with Ball Kicking	1) Double-limb stance, hard surface, ball Kicking, 3 Reps 2) Single- limb stance, hard surface, ball Kicking, 3 Reps 3) Double-limb stance, mini-trampoline, ball Kicking, 3 Reps 4) Single- limb stance, mini-trampoline, ball Kicking, 3 Reps	The participant was instructed to return the ball to the therapist and maintain balance as much as possible after kicking. During this training, the uninjured limb was used for kicking, while the affected limb served as the supporting limb.	If the participant successfully completed 3 repetitions without any errors at each difficulty level.	Errors included A, B, C, and D mentioned earlier.
Single-Limb Hop to Stabilization	1) Target at 18 inch, using arms to aid in stabilizing, 10 Reps 2) Target at 18 inch, hands on hips while stabilizing, 10 Reps 3) Target at 27 inch, using arms to aid in stabilizing, 10 Reps 4) Target at 27 inch, hands on hips while stabilizing, 10 Reps 5) Target at 36 inch, using arms to aid in stabilizing, 10 Reps 6) Target at 36 inch, hands on hips while stabilizing, 10 Reps	The participant was instructed to perform 10 hops in four directions: 1. Anterior-posterior 2. Medial-lateral 3. Anteromedial-posterolateral 4. Anterolateral-posteromedial Three target distances (18, 27, or 36 inches) from the starting point were set. The participant hopped to these targets, stabilized their balance on one limb, then hopped back in the opposite direction to the starting position and stabilized again on one limb.	If the participant successfully completed 10 repetitions without any errors at each level of difficulty.	Errors included A, B, C, and D mentioned earlier.
Hop to Stabilization and reach	1) Target at 18 inch, using arms to aid in stabilizing, 5 Reps 2) Target at 18 inch, hands on hips while stabilizing, 5 Reps 3) Target at 27 inch, using arms to aid in stabilizing, 5 Reps 4) Target at 27 inch, hands on hips while stabilizing, 5 Reps 5) Target at 36 inch, using arms to aid in stabilizing, 5 Reps 6) Target at 36 inch, hands on hips while stabilizing, 5 Reps	The participant hopped, stabilized, and reached back to the starting position. Then, they returned to the starting position and reached to the target position.	If the participant successfully completed 5 repetitions without errors at each difficulty level.	Errors included A, B, C, and D mentioned earlier.

Reps = Repetitions.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics version 23.0 (IBM Corp., Armonk, NY, USA).

The Shapiro–Wilk test was applied to demographic and clinical characteristics, showing that all variables were normally distributed except for age and time since the last ankle sprain. Accordingly, independent t-tests were used for normally distributed variables, the Mann–Whitney U test for non-normal variables, and the Chi-square test for categorical data (gender, affected limb, and dominant limb). No significant baseline differences were found, confirming group comparability. Although the between-group difference in gender distribution did not reach statistical significance (p = 0.067), the near-significant result suggested a potential influence on outcomes; therefore, gender was included as a covariate in subsequent analyses.

For the outcome measures, the Shapiro–Wilk test indicated non-normality for most variables, except ML COP displacement across the three phases of GI and ML COP velocity in the anticipatory phase. Baseline comparisons using independent t-tests or Mann–Whitney U tests, as appropriate, showed no significant differences between groups.

To examine the effects of time, group, and their interaction, a generalized linear mixed model (GLMM) was employed with one within-subject factor (time: pre-intervention, post-intervention, and follow-up) and one between-subjects factor (group: exergaming vs. balance training). A gamma distribution with a log link function was specified for non-normally distributed outcomes, whereas a normal distribution with an identity link was specified for normally distributed variables. Since some outcome variables included negative values due to centering procedures, a constant value was added to all observations to ensure positivity prior to analysis. This linear shift does not alter the relative differences or distributional properties of the data, but allows the application of a Gamma distribution with a log link function, which requires strictly positive values. Where significant interactions were identified, post hoc analyses were conducted within each group to explore simple effects. In the absence of interaction effects, significant main effects were interpreted directly. Effect sizes were reported using partial eta squared (η²) with thresholds defined as small (0.01–0.06), moderate (0.06–0.14), and large (≥0.14). For significant time effects, pairwise comparisons with Bonferroni correction were conducted.

All analyses were performed according to the Intention-to-Treat (ITT) principle. As illustrated in the CONSORT flow diagram, two participants from the intervention group and two from the control group did not complete the one-month follow-up assessment. Missing data were imputed using the Expectation–Maximization (EM) algorithm under the Maximum Likelihood (ML) framework, allowing inclusion of all 34 participants in the final analysis.

Furthermore, to assess the overall impact of the intervention on GI kinetics, effect sizes from multiple gait-related variables were combined. These effect sizes were computed using the standardized mean difference (SMD) for pretest–posttest control group designs, as recommended by Morris (2008), commonly known as dppc2⁴⁹. This method quantifies the difference in pre-to-post changes between groups, standardizes it using the pooled pretest standard deviation, and includes a correction for small sample sizes. By considering both within-group changes over time and between-group differences at post-intervention, it provides a reliable estimate of the true treatment effect⁴⁹. To obtain a single composite effect size for this study, a weighted average of the individual effect sizes was computed. Weights were assigned based on the inverse of the squared standard error (1/SE²), giving greater influence to outcomes with more precise estimates. Positive SMD values indicate a beneficial effect of the intervention, while negative values favor the control group. For outcomes where lower scores reflect better performance (e.g., ML COP displacement during the locomotor phase), values were reversed prior to averaging. This procedure allowed us to summarize the overall effect of the intervention across multiple gait measures.

Results

Descriptive statistics and the results of GLMM are presented in Tables 4 and 5, respectively.

Table 4.

The mean ± standard deviation (SD) of parameters.

Parameters	Phase	Exergaming						Control
		Pre		Post		Follow-Up		Pre		Post		Follow-Up
		Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
COP Displacement [ML] (cm)	1	−1.16	0.54	−1.21	0.75	−1.19	0.97	−1.40	0.96	−1.59	1.12	−1.42	1.04
	2	7.76	2.06	8.28	1.75	8.66	2.65	7.87	2.04	8.03	2.50	7.64	2.57
	3	−2.42	1.38	−1.46	0.97	−2.01	0.91	−2.55	1.84	−1.86	1.26	−1.13	1.24
COP Displacement [AP] (cm)	1	−1.31	0.78	−1.90	1.04	−1.68	1.41	−1.56	1.87	−1.91	0.83	−1.97	1.03
	2	0.30	1.16	1.15	1.93	1.09	1.27	0.11	1.73	1.19	1.87	0.54	1.02
	3	14.57	6.37	15.38	6.59	15.48	6.01	15.48	7.96	17.00	6.88	16.00	5.54
COP Velocity [ML](cm/s)	1	4.26	1.65	4.62	1.26	4.96	2.11	4.83	3.00	3.84	2.42	5.29	2.11
	2	20.73	8.20	21.68	6.82	25.06	11.77	20.86	6.79	21.74	8.27	21.65	8.75
	3	20.54	19.06	26.34	27.48	22.35	17.29	23.21	27.74	24.27	27.72	19.36	17.16
COP Velocity [AP](cm/s)	1	5.53	2.02	7.52	2.78	6.93	3.64	5.63	4.57	4.68	2.02	5.22	3.02
	2	2.77	1.24	4.15	2.66	3.82	2.46	3.28	3.30	4.34	3.93	3.75	2.10
	3	15.42	10.26	16.75	10.24	16.67	21.51	17.44	17.36	10.61	7.42	11.44	10.00

COP = center of pressure; AP = anterior–posterior; ML = medial–lateral.

Phase 1, 2 and 3 are the anticipatory, weight transition and locomotor phase, respectively.

Negative sign in AP direction indicates the backward displacement of COP and in ML direction means displacement towards the step-initiating limb.

Table 5.

Results of the generalized linear model analysis.

Parameters	Phase	Group Effect			Time Effect			Time* Group Effect
		Wald χ2	P	η2	Wald χ2	P	η2	Wald χ2	P	η2
COP Displacement [ML]	1	2.520	0.112	0.026	0.337	0.845	0.004	0.150	0.928	0.002
	2	0.765	0.382	0.008	0.545	0.762	0.006	1.144	0.564	0.012
	3	0.198	0.656	0.002	10.781	0.005*	0.101	4.842	0.089	0.048
COP Displacement [AP]	1	0.326	0.568	0.003	1.702	0.427	0.017	0.163	0.922	0.002
	2	0.682	0.409	0.007	7.873	0.020*	0.076	0.659	0.719	0.007
	3	0.575	0.448	0.006	0.506	0.776	0.005	0.107	0.948	0.001
COP Velocity [ML]	1	0.008	0.929	0.000	3.160	0.206	0.032	2.027	0.363	0.021
	2	0.283	0.595	0.003	1.184	0.553	0.012	0.683	0.711	0.007
	3	0.261	0.610	0.003	1.165	0.558	0.012	5.678	0.058	0.056
COP Velocity [AP]	1	5.315	0.021*	0.053	0.371	0.831	0.004	3.566	0.168	0.036
	2	0.184	0.668	0.002	3.565	0.168	0.036	0.271	0.873	0.003
	3	2.393	0.122	0.024	1.402	0.496	0.014	2.809	0.245	0.028

COP = center of pressure; AP = anterior–posterior; ML = medial–lateral.

Phase 1, 2 and 3 are the anticipatory, weight transition and locomotor phase, respectively.

η2 is effect size (small = 0.01–0.06.01.06, medium = 0.06–0.14.06.14 and large ≥ 0.14).

^*Significant differences (P < 0.05).

Anticipatory phase

Analysis of AP COP velocity during the anticipatory phase revealed a significant between-group difference post-intervention (Wald χ² = 11.067, df = 1, p = 0.001, η² = 0.257, residual df = 32), with the exergaming group exhibiting higher velocities compared to the control group. At the one-month follow-up, the between-group difference was not statistically significant (Wald χ² = 2.020, df = 1, p = 0.155, η² = 0.059, residual df = 32), indicating a small effect size. These findings suggest that the intervention elicited a robust improvement in anticipatory AP COP velocity, which was not maintained at follow-up. All other parameters measured during this phase—including ML COP displacement, ML COP velocity, and AP COP displacement—showed no significant effects of group, time, or group × time interaction (all p > 0.05) (Tables 4 and 5).

Weight transition phase

Analysis of AP COP displacement during the weight transition phase revealed a significant main effect of time (Wald χ² = 7.873, p = 0.020, η² = 0.076), whereas the effects of group (Wald χ² = 0.682, p = 0.409, η² = 0.007) and the group × time interaction (Wald χ² = 0.659, p = 0.719, η² = 0.007) were not significant. Post hoc pairwise comparisons with Bonferroni correction indicated a significant forward displacement of COP (i.e., reduction in backward displacement) from pre-intervention to post-intervention in both groups (Mean Difference = −0.968, 95% CI: −1.810 to −0.127, p = 0.018, η² = 0.024). No significant difference was observed between pre-intervention and follow-up (Mean Difference = −0.607, 95% CI: −1.424 to 0.210, p = 0.226, η² = 0.015), and no significant change occurred between post-intervention and follow-up (Mean Difference = 0.361, 95% CI: −0.516 to 1.239, p = 0.973, η² = 0.004), indicating partial retention of the changes over the follow-up period. Other parameters in this phase—including ML COP displacement, ML COP velocity, and AP COP velocity—were not significantly affected, with no significant effects of group, time, or group × time interaction (all p > 0.05) (Tables 4 and 5).

Locomotor phase

Analysis of ML COP displacement during the locomotor phase revealed a significant main effect of time (Wald χ² = 10.781, p = 0.005, η² = 0.101), whereas the effects of group (Wald χ² = 0.198, p = 0.656, η² = 0.002) and the group × time interaction (Wald χ² = 4.842, p = 0.089, η² = 0.048) were not statistically significant. Post hoc pairwise comparisons with Bonferroni correction indicated a significant reduction in ML COP displacement toward the step-initiating limb from pre-intervention to post-intervention (Mean Difference = −0.822, 95% CI: −1.556 to −0.089, p = 0.022, η² = 0.024) observed in both groups. This change persisted at the one-month follow-up, as demonstrated by a significant reduction from pre-intervention to follow-up (Mean Difference = −0.913, 95% CI: −1.647 to −0.180, p = 0.009, η² = 0.028). No significant difference was observed between post-intervention and follow-up (Mean Difference = −0.091, 95% CI: −0.825 to 0.643, p = 1.000, η² = 0.001), indicating stability of the post-change effect over time. All other parameters in this phase—including AP COP displacement, ML COP velocity, and AP COP velocity—showed no significant effects of group, time, or group × time interaction (all p > 0.05). Additionally, ML and AP COP velocity demonstrated comparable changes across both groups (Tables 4 and 5).

Overall effect

The pooled effect size across GI variables was small but statistically significant (Weighted Mean dppc2 = 0.20, SE = 0.10, 95% CI [0.01, 0.40]). This indicates that, while the magnitude of improvement was modest, the exergaming produced a measurable positive effect on GI performance compared with the control condition, although the wide confidence interval suggests some uncertainty regarding the exact size of the effect.

Discussion

The present study aimed to compare the effects of exergaming and traditional balance training on GI in athletes with CAI. We hypothesized that exergaming would lead to greater improvements in APAs and GI biomechanics compared with balance training. This hypothesis was partially supported. Exergaming resulted in significantly greater improvements in AP COP velocity during the APA phase immediately after training, indicating enhanced anticipatory control; however, this effect was not retained at the one-month follow-up. During the execution-oriented phases of GI, including weight transition and locomotor phases, both interventions produced comparable improvements. Specifically, AP COP displacement increased during the weight transition phase, indicating more effective forward progression. In the locomotor phase, ML COP displacement toward the swing limb was reduced, reflecting improved lateral stability. These changes were partially to fully retained over time. Overall, when GI parameters were pooled, exergaming showed a small but statistically significant advantage over balance training.

APAs rely on feed-forward neural mechanisms originating from cortical and subcortical centers, particularly the supplementary motor area (SMA) and basal ganglia, which pre-activate postural muscles in preparation for movement initiation^35,39,50. In individuals with CAI, impaired preparatory brain activity has been linked to altered COP trajectories during the APA phase of GI, reflecting deficient anticipatory control³⁹. Importantly, targeted interventions have been shown to enhance cortical motor preparation, leading to shorter APA duration, improved perceived ankle stability, and more efficient postural adjustments during GI⁴⁷. Evidence from older adults further supports the notion that exergaming can strengthen cortical motor preparation and facilitate GI biomechanics^13,25. Within this framework, the greater improvements in AP COP velocity during the APA phase observed after exergaming in the present study likely reflect enhanced feed-forward control. However, the absence of sustained effects at the one-month follow-up suggests that these neural adaptations may be transient if not adequately reinforced. Incorporating booster sessions, increasing training dose, or combining exergaming with complementary neuromuscular or balance-oriented exercises may help consolidate neural plasticity and improve retention of anticipatory control gains.

In addition, exergaming imposes substantial multisensory and cognitive demands. The continuous integration of visual, proprioceptive, and auditory inputs within a virtual environment promotes sensorimotor integration and heightened neural engagement, supporting more effective motor planning during the APA phase¹³. Moreover, many exergames necessitate sustained attention and rapid responses to visual stimuli. For instance, Table Tilt and Penguin Slide require planning, while Soccer Heading involves decision-making and response inhibition^23,46. Collectively, these games integrate both cognitive and motor demands, fostering anticipatory control through enhanced neural engagement. Consistent with this interpretation, improvements in reaction time and neurocognitive performance have been reported in individuals with CAI following exergaming interventions⁵¹.

Additionally, games such as Soccer Heading, Snowboard Slalom, and Tightrope Walk require players to anticipate movement timing, predict trajectories, and navigate obstacles, thereby reinforcing neural processes involved in predictive postural control. Although neural activity was not directly measured in the present study, prior neuroimaging research has demonstrated reorganization within prefrontal and motor cortices following video game training, along with activation of mirror neuron networks through movement observation and visual–motor integration^13,52,53. These mechanisms may collectively contribute to the observed enhancements in APA performance.

The postural demands inherent in virtual reality (VR)–based exercises provide an additional explanation for improved anticipatory control. Exergaming typically involves continuous multidirectional weight shifting across varying speeds and amplitudes, engaging coordinated ankle, hip, and trunk strategies that are essential for generating appropriate COP trajectories during GI²⁵. Repeated exposure to these demands likely promotes motor learning and refinement of stored movement programs. Furthermore, perturbation-oriented tasks embedded within exergames—such as single-limb stance combined with contralateral limb movement—introduce internal disturbances that challenge postural stability and enhance neuromuscular control⁸. Supporting evidence from EMG studies demonstrates earlier muscle activation onset following VR-based training across different clinical populations, reinforcing the role of exergaming in optimizing anticipatory neuromuscular strategies^8,54–56.

In contrast to the anticipatory phase, both exergaming and traditional balance training produced comparable improvements during the execution-oriented phases of GI, including weight transition and locomotor phases. During weight transition, the observed forward shift of the COP indicates a more effective anterior progression of the COM, reducing braking forces and facilitating smoother momentum transfer prior to stepping. In the locomotor phase, reductions in ML COP displacement toward the swing limb suggest improved lateral stability and step efficiency. Notably, these execution-related adaptations were largely retained at follow-up. Neurophysiological evidence supports a partial dissociation between anticipatory and execution processes^57–60. APAs are primarily governed by feed-forward mechanisms mediated through cortical–subcortical circuits, whereas execution phases following movement onset rely more heavily on feedback-driven control involving brainstem and spinal pathways⁶¹. Within this framework, the present findings suggest that exergaming preferentially enhances feed-forward neural strategies underlying APAs, while both interventions similarly strengthen feedback-based mechanisms responsible for movement execution.

Proprioception plays a central role in both control modes. In feed-forward regulation, proprioceptive input contributes to updating internal models of body dynamics, whereas in feedback control it provides real-time information necessary for rapid error detection and corrective responses^39,62–64. The balance training protocol in this study—including single-limb stance, hop-to-stabilization, and perturbation-based tasks—likely enhanced proprioceptive processing and reinforced feedback-based mechanisms, thereby improving stability during the execution phase. Both groups appeared to benefit from these proprioceptive gains, which likely explains the similar improvements observed in execution-oriented GI measures. Supporting evidence indicates that both balance exercises and exergaming can effectively enhance proprioception in individuals with CAI^65,66. Moreover, improvements in both groups may be attributed to increased muscle strength and more efficient muscle activation patterns. Balance training has been shown to optimize muscle activity, reduce activation delays, and strengthen ankle muscles in individuals with CAI^67–70. Similarly, VR-based exercises have been reported to improve plantar flexor strength⁷⁰.

While exergaming has demonstrated benefits for balance, APAs, and task-specific motor control, its superiority over conventional training remains inconclusive. Most studies, including the present one, report greater improvements with exergaming^26–31, whereas a few found comparable outcomes^8,32, with conventional training occasionally showing advantages in muscle activation and postural strategies⁸. These discrepancies likely arise from methodological heterogeneity. Intervention design is different across studies, most trials employed the Nintendo Wii Fit platform^8,26,29–32, while others used alternative systems such as the Stepmania dance mat²⁷ or motion capture with D-flow²⁸. Intervention “dose” varied considerably, ranging from short protocols of eight sessions over four weeks²⁸ to intensive regimens of 36 sessions across three months³¹. Comparator conditions also differed, with some studies using only balance training and others incorporating strengthening exercises^26,29–31. The types and number of games also varied, ranging from broad multidimensional protocols with 11 games²⁶ to more restricted approaches with only two games²⁷. Assessment tools also varied across studies, including force plates³², EMG for APAs and CPAs⁸, Biodex Balance System^30,31, and clinical tests such as SEBT and YBT^8,26–29. Follow-up assessments were limited, with only two studies including one-month retention tests^29,32. Although the studies differed in methodology, the primary factor underlying the divergent outcomes may be the specific tasks employed and the postural outcomes extracted from them. In particular, the two studies reporting different results involved high-demand tasks, such as jump-landing and ball kicking, which likely influenced the types of postural metrics assessed. This interpretation is supported by the observation that studies using similar low- to moderate-demand tasks generally reported consistent outcomes. Thus, it appears that task characteristics and the corresponding outcome measures may play a more prominent role than other methodological factors in explaining the observed inconsistencies.

Limitations

Future studies could benefit from integrating muscle activity assessment via EMG with brain activity recordings—using EEG or functional near-infrared spectroscopy (fNIRS)—and kinetic measures to provide deeper insights into the underlying mechanisms. Another limitation of the present study is the use of the Wii Balance Board as the primary training tool. Although this device is practical and accessible, it may not adequately capture the dynamic and complex demands of real-life sports. The structured environment of exergaming might also limit the ecological validity of training outcomes, as athletes are generally exposed to complex and unpredictable movements that are difficult to replicate in laboratory-based tasks. To improve the transferability of training effects, future research could integrate more sport-specific exercises or employ advanced VR–based systems that better simulate real-game conditions. Moreover, another limitation of the study is the potential influence of game order on motor and cognitive engagement. Although the order of games was determined based on participants’ preferences, this sequencing may have introduced variability in exposure across participants, which could affect the comparability of engagement levels. The study sample consisted exclusively of recreational athletes, which may limit the generalizability of the findings to competitive athletes or non-athletic populations. Recreational athletes differ from elite athletes in terms of training intensity, neuromuscular demands, and exposure to complex sport-specific tasks, which may affect the transferability of the intervention to high-performance settings. In contrast, non-athletic individuals may present lower baseline neuromuscular control, reduced strength, and different motor learning capacities, potentially leading to different magnitudes or patterns of adaptation to the intervention. Therefore, caution is warranted when extrapolating these findings beyond recreational athletes. Future research should investigate whether similar effects are observed in competitive athletes and non-athletic populations with varying levels of physical activity and functional demands.

Conclusion

Exergaming produced a distinct phase-specific benefit in athletes with CAI by enhancing AP COP velocity after training, reflecting more efficient anticipatory postural adjustments; however, this effect was not sustained at follow-up. Both exergaming and balance training contributed to improvements during the execution-oriented phases of GI, including weight transition and locomotor control, with effects that were largely retained over time. Overall, although the pooled effect size was small, exergaming yielded gains beyond those achieved with balance training.

Abbreviations

ANOVA

Analysis of variance

AP

Anterior–posterior

APA

Anticipatory postural adjustment

CAI

Chronic ankle instability

CAIT

Cumberland ankle instability tool

CNS

Central nervous system

COG

Center of gravity

COM

Center of mass

COP

Center of pressure

CPA

Compensatory postural adjustment

EEG

Electroencephalography

EM

Expectation–maximization

EMG

Electromyography

fNIRS

Functional near-infrared Spectroscopy

GI

Gait initiation

GLMM

Generalized linear mixed model

GRF

Ground reaction force

ITT

Intention-to-treat

L1

Landmark 1

L2

Landmark 2

LAS

Lateral ankle sprain

ML

Medio-lateral

ML (in algorithm)

Maximum likelihood

SMA

Supplementary motor area

SMD

Standardized mean difference

SEBT

Star excursion balance test

VR

Virtual reality

YBT

Y balance test

Author contributions

R. Kh: Conceptualization, Supervision, Methodology, Software, Validation, Writing – Review & Editing. S.S.S: Data Curation, Formal Analysis, Investigation, Visualization, Writing – Original Draft. Both authors read and approved the final version of the manuscript.

Funding

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

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics

All participants provided written informed consent prior to enrollment. The study protocol was reviewed and approved by the Ethics Committee of Tehran University of Medical Sciences (Approval code: IR.TUMS.FNM.REC.1401.146). All methods were performed in accordance with the relevant guidelines and regulations.