Twelve active male university students (age: (mean ± SD) 26.8 ± 2.3 years old; height: 171.8 ± 4.6 cm; body weight: 67.0 ± 9.1 kg) with 6.0 ± 4.2 years of training experience were recruited, including 8 badminton, 2 basketball, 2 fitness and running specialists. The predicted recruitment sample size was calculated by using G*Power software, a large effect size of 0.8 was set for repeated measures analysis of variance (ANOVA), at least 10 participants were required with an α err probability of 0.05, and power level of 0.8 (Shi QQ. et al., 2022a; Shi Q-Q. et al., 2022b). The exclusion criteria included: (1) the presence of any injury, illness, or medical condition identified via the Physical Activity Readiness Questionnaire (PAR-Q), and (2) failure to meet the physical training requirements, (i.e., ≥ 2 years of training history, or ≤ 1 h/session with 2 sessions/week in the past 2 years). They were asked to sign a consent form prior to the experiment after given a briefing on the study. Human subject ethics approval for this study was granted by the Institutional Review Board of the University Ethics Committee (HSEARS20240206005). Each participant visited the laboratory once for testing. They were required to refrain from drinking caffeine or taking stimulants, and no physical training for 24 h before they participated in the tests.

Adjustable 3D polyurethane insoles (CPWGSM, United States of America) with a prefabricated arch-support that has three levels of stiffness, including soft, semi-rigid, and rigid, are used in this study (see Figure 1). The information for the stiffness level is provided by the manufacturer. Due to the variations in stiffness, the 3D arch-support insoles have an average difference of 3.75% (1 mm) in arch height and 2% (0.95 g) in weight. The participants wore a commonly available type of training shoe (Double Star, China) in their respective foot size with each type of 3D arch-support insole or without any insole (control). To minimize the order effect, conditions were randomized using a balanced Latin square design (Bradley, 1958), ensuring each stiffness level appeared equally in each position across participants. Additionally, sufficient rest intervals (≥15 min) were enforced between trials to mitigate fatigue or carryover effects.

Prior to the data collection, the participants completed a 10-min warm-up session (LaPorta et al., 2013) and acclimatized to the testing conditions by practicing 3–5 trials of CMJs. Then a total of 16 reflective markers of 14 mm in diameter were placed on the lower limbs of the participants, according to the Vicon Plug-in Gait marker set protocols (Shi Q-Q. et al., 2022b; Kadaba et al., 1989). To ensure consistency and reliability, the same senior researcher attached the markers onto each participant.

During the CMJ trials, the participants would stand on the centre of the force plates with their hands on their iliac crest to minimize arm swing (Kajiwara et al., 2019). They were instructed to perform a CMJ and land back onto the center of the force plates, and maintain a neutral and quasi-static state for 3 s to ensure that the data were properly captured (Wikstrom et al., 2005). Trials were considered unsuccessful if the participant failed to remain stationary for the required 3 s or removed his hands from his iliac crest during the CMJ. To prevent fatigue, the participants were given a break of 30 s between trials, and a 15-min break was also given between the different orthosis conditions. Kinematic and kinetic data were collected by using two synchronized force plates with dimensions of 60 cm × 40 cm (AMTI, United States of America) and an eight-camera motion analysis system (VICON, Nexus 2.0 Inc., Oxford, United Kingdom) at a sampling frequency of 200 Hz. Five successful trials were recorded for each condition.

Based on the existing literature that have analyzed data on CMJs, the phases of the CMJs are defined according to six time points (McMahon et al., 2018) in Figure 2 (black curve indicates the vGRF during a CMJ (total time is 100%), and red curve indicates the velocity of the center of mass (vCOM) during a CMJ (total time is 100%)): (1) weighing phase, which is from T0 to T1, where T1 is the time when the vGRF is reduced by a threshold equal to 5 times the standard deviation of the initial weighing value; (2) unweighting phase, which is from T1 to T2, where T2 is the time when the vCOM meets the minimum value; (3) braking phase, which is from T2 to T3, where T3 is the time when vCOM increases to over 0 after T2; (4) propulsion phase, which is from T3 to T4, where T4 is the time when vGRF decreases to less than 10% of the initial weighing value; (5) flight phase, which is from T4 to T5 where T5 is the time when vGRF increases to greater than 10% of the initial weighing value; and (6) landing phase, which is from T5 to T6 where T6 is the time when the vGRF decreases to 95% of the initial weighing value.

The kinematics of the participants was captured with the reflective markers. The ankle, knee, and hip joint angular displacements were calculated for each participant and compared based on with and without the different arch-support conditions when wearing neutral shoes. The kinetics data were calculated by using the vGRF during jumping. The ISB global joint coordinate system used for the calculations is presented in Figure 3. Joint angles, forces, and moments are expressed in three anatomical planes, including the sagittal, frontal, and transverse planes. Following the Plug-in Gait Reference Guide (Vicon Motion Systems Limited) (Plug‐in Gait Reference Guide, 2023), Supplementary Table S1 presents the biomechanical variables for the lower limb joints which were investigated on the three planes during CMJ cycles.

The three best jumps for each participant were selected based on maximal jump height, with proper landing (i.e., full foot contact and no balance loss). These trials were obtained through marker labeling, gap filling, and establishing a skeletal model of lower limbs to determine the kinematic and kinetic information in the Vicon software. All of the joint center trajectories were smoothed by using a recursive fourth-order low-pass Butterworth filter, with a cutoff frequency of 6 Hz selected through a residual analysis and visual inspection of the data (McErlain-Naylor et al., 2014). The moment and force parameters were normalized to the height and weight of the participants (Shi Q-Q. et al., 2022b). The dominant foot was determined by conducting a procedure during which the participant kicked a ball (Shi QQ. et al., 2022a). All of the participants used their right foot as their dominant foot.

SPM analyzed the effect of insole condition (single factor: soft/semi-rigid/rigid/no 3D arch-support insole) on time-normalized biomechanical trajectories. Unlike factorial ANOVA, SPM does not discretize time into phases but instead identifies supra-threshold temporal regions where conditions differ significantly, accounting for temporal smoothness via random field theory (Pataky, 2010). Initially, the one-way repeated measures analysis of variance (ANOVA) was applied to the normalized time series to detect the overall differences among the experimental insole conditions. Where significant effects were identified (p < 0.05), post hoc paired t-tests were performed over the normalized time series to localize the differences between the specific orthotic configurations. Technical procedures were implemented in MATLAB (MathWorks Inc., United States of America) by using the spm1D package (v0.4) (Patak and y, 2012), which addresses the multiple-comparison problem inherent to time-series analyses. Further methodological details can be found in previous studies (Pataky, 2010; Penny et al., 2011).

The SPM analysis followed four sequential steps.

a. The test statistics (F- or t-values) were calculated at each time point within the normalized time series.

Our findings showed that the 3D arch-support insoles significantly altered the ankle biomechanics during propulsion-flight phase of CMJ. The results are consistent with our hypothesis that some impactful changes failed to be detected by previous zero-dimensional data analysis. Specifically, the insole led to significantly reduced ankle plantarflexion at 59%–69% of CMJ, and increased ankle adduction at 58%–76% of CMJ compared to the control condition. The rigid 3D insole significantly decreased ankle plantarflexion at 65%–66% (propulsion), and increased ankle adduction at propulsion (66%–67%). The semi-rigid 3D insole significantly increased ankle adduction at propulsion-flight (63%–72%). These narrow time windows likely correspond to critical transition points in force generation, where the foot shifts from mid-stance to push-off, suggesting that 3D arch-support insoles most strongly modulate ankle mechanics during rapid force redirection. While brief, these intervals may reflect key instants where foot rigidity influences propulsion efficiency.

These kinematic changes aligned with functional role of ankle plantarflexion and adduction dominate the propulsive phase (Welte, 2020). The observed increase in ankle adduction may be attributed to the medial support provided by the insoles, which modifies frontal-plane motion. This finding partially agrees with Yu et al. (2007), who also reported increased ankle adduction with arch-support interventions during landing. Aligning with Ho et al. (2019) reported that no significant interactions between insoles and foot type for jump height, kinematic and kinetic variables in basketball players, but the ankle eversion was significantly reduced by using prefabricated insoles during CMJs take-off moment, suggesting that the medial support of 3D arch-support insoles primarily altered the frontal plane kinematics.

The windlass effect is a mechanism in the foot that occurs during the push-off phase of jumping, may explain the altered ankle kinematics. Ker et al. (1987) proved that the elastic properties of human’s foot arch is essential for windlass effect. During dynamic loading, the foot arch stores strain energy, facilitating the windlass effect. This energy-releasing mechanism complements the elastic action of the Achilles tendon during push-off (Ker et al., 1987). The 3D arch-support insoles could enhance this mechanism by reinforcing the medial longitudinal arch, thereby facilitating more effective force transfer during propulsion (Welte et al., 2018). Additionally, the semi-rigid insole’s faster compression (Figure 4) could optimize elastic recoil during weighting-unweighting (21%–24%), mirroring the energy-storing role of the arch described by Ker et al. (1987). This suggests that the insoles augment the foot’s natural spring-like function, facilitating the foot’s capability in storing strain energy and returning it in as elastic recoil, whilst improving push-off efficiency.

While the 3D arch-support insoles did not significantly affect knee and hip joint moments, they did reduce hip flexion and abduction moments during the braking-propulsion phase compared to the control. Excessive hip abduction can cause undesirable lateral shifts in center of pressure (Lee and Powers, 2014). The insoles appear to influence hip stability in both sagittal and frontal planes during this critical push-off phase by significantly modifying ankle biomechanics. Our study is the first to report increased ankle adduction during propulsion–flight with semi-rigid 3D insoles by SPM. This finding suggested that 3D insoles, especially the semi-rigid one, may optimize push-off efficiency by promoting faster energy return and reducing unnecessary motion. Future research should explore whether these biomechanical adjustments translate to improved jump performance (e.g., height or power) in athletic populations.

Following the ankle kinematic changes, most of significant ankle kinetic alteration occurred during landing (90%–100%) by wearing 3D arch-support insoles. The rigid insole significantly reduced transverse ankle internal rotation moment during 91%–100% of CMJ, with reduced ankle anterior force at 90%–91% and ankle medial force during 94%–100%. Semi-rigid 3D arch-support insole significantly increased ankle adduction during early landing (87%–91%), with reduced ankle internal rotation moment during 97%–100% of CMJ. While soft insole significantly reduced ankle anterior force during 90%–100%. The results verified our previous hypothesis that significant ankle biomechanical changes occurred during landing, but not at the peak value moments.

Notably, excessive moments of internal rotation in the transverse plane have been reported to exacerbate anterior cruciate ligament strain (Lee and Shin, 2021). This reduction of the internal ankle rotation moments aligns that arch’s ligaments (e.g., plantar calcaneonavicular/spring ligament) resist flattening under load, limiting aberrant joint motion (Ker et al., 1987). The rigid and semi-rigid insoles may amplify this stabilizing effect, mitigating forces linked to anterior cruciate ligament strain (Lee and Powers, 2014) by reinforcing midfoot rigidity. Previous study claimed that the lateral and anterior forces peak earlier which is considered to be less functional stability (Caulfield and Garrett, 2004). The rigid 3D arch-support insole may enhance ankle joint stability in the frontal and transverse planes during the propulsion and landing phases (Figure 7). We found that the insoles have little effect on the vGRF, which aligns with the findings in Wang et al. (2020). Our finding is consistent with the understanding that stiffer arch-support insoles increase midfoot or rearfoot rigidity without substantially altering the vGRF during landing movement (Zhang et al., 2012). Differently from previous study, Lou et al. (2015) whose participants showed a significantly reduced peak moment of the internal ankle rotation with the use of arch-support insoles. Our study found that 3D insoles significantly influenced the ankle biomechanics from mid-landing until stabilization.

These observed biomechanical changes in the ankle joints are largely due to the changes in the alignment of the foot with the use of the rigid and semi-rigid 3D arch-support insoles in this study. These 3D insoles made with varying material stiffness are recommended to mainly redistribute plantar forces to offer shock absorption and better cushioning (Shi QQ. et al., 2022a; Shi Q-Q. et al., 2022b). The insoles’ materials (i.e., rigid, and semi-rigid) may complement this by augmenting energy dissipation during landing, as evidenced by reduced ankle anterior forces (90%–100%). This synergy between intrinsic (arch) and extrinsic (insole) mechanisms could explain the protective effects observed (Ker et al., 1987). For clinicians, arch-support insoles are a vital tool for managing painful foot and lower extremity pathologies, which prevent injuries, and optimize the biomechanics during sports and other weight-bearing activities (Werd et al., 2010). The rigid and semi-rigid 3D insoles effectively influenced ankle joint in the frontal and transverse planes during landing phases, which is crucial for maintaining balance and reduce impact forces during landing, thereby protecting against over-use injuries. In line with Alirezaei Noghondar and Bressel (2017), our findings suggest that the 3D insoles especially the rigid one modulate loading pattern, thus potentially reducing the risk of injury during jumping-landing tasks. However, very few studies have investigated the biomechanical changes by examining the different degrees of rigidity of 3D arch-support insoles on CMJs, so CMJ performance by using 3D insoles with different degrees of stiffness warrants further investigation.

There are several limitations in this study. First, while the 3D arch-support insoles were designed to influence medial longitudinal arch mechanics, foot structure (e.g., arch height, ligamentous stiffness) was not assessed. Since the insoles’ effects may differ based on intrinsic foot stiffness or pre-existing arch morphology (e.g., pes planus vs. pes cavus), without quantifying these anatomical factors, the generalizability of our findings across foot types remains uncertain. Future studies should incorporate foot structure assessments (e.g., arch index, 3D scanning, or dynamic arch kinematics) to determine how insole efficacy interacts with individual anatomical variations. Second, this study focused on kinematic and kinetic outcomes but did not examine the interaction between jump performance (e.g., height, power) and biomechanical changes. A comprehensive analysis linking these variables could clarify whether the observed ankle adduction or reduced internal rotation moments translate to functional improvements (e.g., energy return, injury resilience). Finally, the insoles’ material properties (rigid vs. semi-rigid) were standardized, but their interaction with the foot’s natural compliance (e.g., plantar aponeurosis tension) was not quantified. This limits mechanistic interpretations of how extrinsic (insole) and intrinsic (arch) stiffness jointly modulate biomechanics, which shall be investigated in future work.