An a priori power analysis (G*Power 3.1) indicated that 26 participants were required to achieve 0.95 statistical power at α = 0.05, based on a previous study's group-by-time interaction effect size (η²_p = 0.122 equals to effects size f = 0.372) for CoP_RMS during single-leg stance in individuals with CAI undergoing neuromuscular electrical stimulation (pre: 8.13 ± 1.07 mm vs. post: 6.60 ± 1.14 mm) (37).

Participants were recruited from August to October 2024 via university e-newsletters, posters, and direct emails. Seventy-five individuals were screened for eligibility using International Ankle Consortium guidelines and additional criteria (38), with 40 providing informed consent. Inclusion criteria were: (1) history of ≥1 ankle sprain >1 year prior with acute symptoms (pain, swelling, activity limitation >1 day); (2) age 18–25 years without athletic specialization; (3) ≥ 2 episodes of ankle instability/"giving way” in the past 6 months; (4) persistent instability/functional impairment; and (5) Cumberland Ankle Instability Tool score <24 (39). Exclusion criteria included lower-limb fractures/surgeries, acute injury within 3 months, bilateral CAI, or neurological disorders impairing motor control (e.g., cerebellar disorders, stroke) (33). The study was approved by the Shandong Sport University Ethics Committee (No. 2023036) and adhered to the Declaration of Helsinki.

This single-blind RCT employed a computer-generated random sequence to allocate 40 participants (1:1) into two interventions: (1) HD-tDCS + BBT and (2) BBT (sham HD-tDCS + BBT). Both interventions underwent six weeks of intervention (3 sessions/week, 20 min/session), with HD-tDCS/sham and BBT administered concurrently. The protocol comprised: 10-minute warm-up, four targeted exercises (30-second each, 30-second rest intervals after each exercise, cycle repeated five times), totaling 20-minute of exercise, followed by 10-minute cooldown. Static and dynamic postural stability were assessed pre- and post-intervention, with test sequences randomized via computer-generation to minimize order effects.

HD-tDCS was delivered via a StarStim8 device (Neuroelectrics, Spain) using a 10/20 EEG-compliant montage of five 5-mm electrodes: one anode (Cz) and four cathodes (Fz, Pz, C3, C4) (40) (Figure 1). Active stimulation applied 2 mA to the anode, with return current distributed across cathodes. The protocol included a 30-second ramp-up to 2 mA, 19-minute at 2 mA, and a 30-second ramp-down. Sham stimulation mirrored this timing but delivered subthreshold currents (<0.1 mA) during the 19-minute phase to preserve blinding (41). Neuro-modeling confirmed focal targeting of sensorimotor cortices (M1/S1) corresponding to foot-ankle representations (42).

Under the guidance of certified instructors, participants performed Bosu ball training following a progressively intensified program. The intervention involved standing on the soft surface of the Bosu ball with the affected limb positioned superiorly and the unaffected limb adjacent. During weeks 1–2, the training progression consisted of: single-leg stance maintenance; single-leg stance with lower extremity anteroposterior swing (30°–45°); single-leg stance with lower extremity mediolateral swing (20°–30°); and single-leg squats. During weeks 3–4, the training progression consisted of: swallow balance positions; single-leg stance with anteroposterior swing (45°–60°); single-leg stance with lower extremity mediolateral swing(30°–45°); and dynamic single-leg squat take-ups. During weeks 5–6, the training progression consisted of: single-leg stance with ball-catching; single-leg stance with lower extremity anteroposterior mediolateral swing(45°–60°); single-leg stance with lower extremity mediolateral swing (30°–45°); and functional reaching tasks involving forward trunk flexion to touch edge of Bosu ball while maintaining single-leg stability(Figure 2).

Participants completed the single leg stance task to assess static postural stability. After reviewing procedures, warming up, and practicing (≥3 trials), they stood on their affected leg atop a force platform (AMTI, Watertown, MA, USA), hands on hips, gaze fixed forward. The unaffected leg was raised to calf level, with the affected foot maintaining full contact for 30 s. Trials were discarded and repeated if: (1) limbs made contact, (2) hands moved from hips, or (3) trunk/hip deviation exceeded 30. Three valid trials were averaged for analysis, with ≥1-minute rest between attempts to minimize fatigue.

Participants performed a drop-landing task to assess dynamic postural stability (43). Standing on a 20 cm wooden platform in front of a force plate, they positioned feet shoulder-width apart, hands at their waist, and gaze fixed forward. Following instructions, participants stepped forward with their affected limb, dropped onto the force plate, and stabilized on the affected leg for 5 s (Figure 3) for three trials. A successful trial required landing without losing balance or corrective movements. Prior to the formal testing, participants completed three practice trials to become familiar with the procedure.

During static postural stability test, CoP_RMS was calculated from anteroposterior (AP) and mediolateral (ML) directions using CoP data sampled at 1,000 Hz. The raw data were filtered using a fourth-order low-pass Butterworth filter with a 12 Hz cutoff frequency (14). Filtered data were used to compute CoP_RMS (mm) for each participant using the following formulas (14):(1)CoP_RMSap=∑(xi−x¯)2N−1(2)CoP_RMSml=∑(yi−y¯)2N−1where x_i and y_i represent CoP coordinates in AP and ML directions, while x bar and y bar denote their means. The denominator N−1 reflects sample-based calculation.

During dynamic postural stability test, ground reaction force (GRF) data were recorded at 1,000 Hz and filtered using a fourth-order low-pass Butterworth filter with a 12 Hz cutoff frequency (14). Filtered data from initial landing (GRF > 10 N) to 5 s post-landing were used to compute time to stabilization (TTS) through sequential average using the following formulas (44):(3)SequentialAverageTTSap(n)=∑n=11000Fx/n(4)SequentialAverageTTSml(n)=∑n=11000Fy/nwhere Fx and Fy represent AP and ML GRF components. TTS was defined as the time from landing to when the sequential average of each component remained within ±25% of the standard deviation of the overall mean GRF for ≥1 s (45) (Figure 4).

Normality was confirmed using the Shapiro–Wilk test. A two-way mixed-design ANOVA evaluated main effects and interactions. Group (HD-tDCS + BBT vs. BBT) was specified as the between-subjects factor, and time (week0 vs. week7) as the within-subjects factor. Significant interactions were decomposed using Bonferroni-adjusted post hoc pairwise comparisons with correction for multiple testing. Effect sizes were reported as partial eta squared (η²ₚ: small = 0.01–0.06, moderate = 0.06–0.14, large > 0.14) for ANOVA results (46) and Cohen's d (trivial < 0.20, small = 0.21–0.50, medium = 0.51–0.80, large > 0.81) (47) for post hoc contrasts. Data are presented as mean ± standard deviation (SD). Significance was set at p < 0.05.

This study demonstrates that both active and sham HD-tDCS, when combined with BBT, significantly improved static and dynamic postural stability in individuals with CAI in both AP and ML directions, underscoring the effectiveness of BBT as a rehabilitative intervention. These findings align with prior research: one study reported that unstable surface training enhances postural stability compared to stable surfaces by increasing neuromuscular demands and sensory integration (32), while another showed that such training elevates muscle activation and proprioceptive feedback to optimize joint stabilization (48). The observed improvements in postural stability following BBT in CAI may stem from its dual mechanisms of enhanced sensory input and neuromuscular adaptation. The compliant surface of the Bosu ball increases sensory stimulation by altering foot-support contact and pressure distribution, which amplifies proprioceptive input (49, 50). A meta-analysis confirmed that augmented sensory input significantly enhances postural stability in CAI populations (51), aligning with evidence of a strong correlation between proprioceptive acuity and postural control in this cohort (52). The inherent instability of the Bosu ball introduces controlled postural perturbations, promoting sensory reweighting—a CNS process that recalibrates reliance on visual, vestibular, and somatosensory inputs to compensate for instability (53). This adaptive mechanism synergizes with neuromuscular demands, as maintaining balance on an unstable surface requires dynamic adjustments to the center of gravity within functional limits (54, 55), thereby counteracting CAI-related deficits in lower limb strength and postural instability (56). Notably, a 4-week unstable surface training protocol improved both muscular strength and postural stability in CAI (48), supporting the results that compliant surfaces elevate neuromuscular demands, fostering enhanced muscle activation and force generation (57). Consequently, these neuromuscular adaptations contribute to improved postural stability in individuals with CAI.

Our findings demonstrate that HD-tDCS combined with BBT exhibited superior efficacy compared to BBT alone in enhancing static and dynamic postural stability among individuals with CAI, these results align with previous studies showing that HD-tDCS paired with foot-core exercises improves proprioception and static balance in healthy adults (36), and that targeting M1/S1 with HD-tDCS during short-foot exercises enhances proprioception and dynamic balance in CAI populations (42). Additionally, evidence that anodal tDCS over M1, when coupled with targeted muscular attention, augments motor cortex plasticity—evidenced by increased motor evoked potentials and reduced short-interval intracortical inhibition—further supports the synergistic effects of HD-tDCS and sensorimotor training on postural stability and motor learning (58, 59).

The superior efficacy of the combined intervention may be attributed to enhanced somatosensory integration, facilitated by HD-tDCS-induced neuromodulation of sensorimotor networks. Primary, HD-tDCS over S1 and M1 likely optimizes cortical excitability, improving sensory processing and motor output during BBT. This aligns with prior work demonstrating that tDCS enhances peripheral somatosensory acuity—evidenced by reduced vibration detection thresholds at the plantar surface and improved hallux sensitivity—thereby refining foot-ankle sensorimotor integration and postural control (60, 61). Such effects may stem from tDCS-mediated modulation of S1 excitability, which could synergize with proprioceptive training to enhance dynamic stability (62). Furthermore, tDCS applied over adjacent temporal-parietal regions has shown benefits for vestibulo-perceptual function (63), suggesting that stimulation effects may extend beyond targeted areas (M1/S1) to interconnected cortical networks involved in multisensory integration, collectively contributing to improved postural outcomes. Secondary, HD-tDCS may enhance postural stability via M1-mediated modulation of lower limb motor output. Individuals with CAI exhibit reduced M1 excitability projecting to the peroneus longus compared to controls (57, 64). Anodal tDCS increases M1 excitability by decreasing resting membrane potential in targeted regions, thereby augmenting corticospinal drive (34). This neuromodulation persists post-stimulation, reducing short-interval intracortical inhibition and enhancing voluntary muscle activation, which strengthens peroneal muscle contributions to postural control (59, 65).

Our results indicate that the additional benefits of HD-tDCS occur specifically in the ML direction. This finding is partially supported by a previous study, which demonstrated that compared to a 4-week foot core training program, HD-tDCS improved passive kinesthesia thresholds for ankle inversion and eversion in healthy individuals, but had limited effects on ankle proprioception for plantarflexion and dorsiflexion (36). Neuromuscular control systems exhibit direction-dependent modulation in postural compensation responses. During perturbations in AP direction, postural stability is maintained through coordinated limb swing patterns and compensatory foot displacement (66). Conversely, perturbations in the ML direction may present greater neuromuscular challenges due to anatomical constraints in lateral limb repositioning (66). The stabilization response in the ML direction initiates with activation of the ankle eversion muscles to counteract inversion stresses (67, 68), a mechanism potentially compromised in individuals with CAI, thereby increasing the risk of sprain recurrence. This directional situation shows a higher correlation between ML stability deficits and fall risk compared to AP instability (69).