Vibration-Based Recovery Interventions Improve Perceived Fatigue, Blood Lactate Clearance, and Isokinetic Muscle Function Following Exercise-Induced Fatigue in Amateur Swimmers

Abstract

High-intensity or repetitive exercise induces metabolic stress and neuromuscular fatigue in skeletal muscle. Using a within-subjects repeated-measures crossover design, eight male amateur swimmers completed five experimental sessions at one-week intervals. Following an isokinetic fatigue protocol, five recovery interventions were applied in a randomized order: control (NT), foam roller (FR), vibration foam roller (VFR), and whole-body vibration at 12 Hz (WBV-12) and 20 Hz (WBV-20). The isokinetic fatigue protocol produced a significant reduction in bilateral extensor peak torque (229.2 ± 37.8%BW to 189.8 ± 27.5%BW; t(7) = 4.19, p = 0.004, d = 1.48), confirming successful fatigue induction. Outcome measures included visual analog scale (VAS) scores, blood lactate concentration, and knee extensor/flexor peak torque (%BW) assessed at three time points. A two-way repeated-measures ANOVA (intervention × time) revealed significant main effects of recovery methods at the post-recovery time point for VAS scores (F(4,28) = 5.98, p = 0.001, η²g = 0.248), blood lactate (F(4,28) = 5.12, p = 0.003, η²g = 0.226), and isokinetic peak torque (F(4,28) = 10.75, p < 0.001, η²g = 0.226). Post hoc Bonferroni analysis indicated that VFR and WBV-20 produced significantly higher lactate recovery rates than NT. Active recovery interventions produced lower perceived fatigue scores and greater lactate reductions than passive rest; however, individual Bonferroni pairwise comparisons for VAS and blood lactate did not reach adjusted significance, and these findings should be considered preliminary. WBV-20 demonstrated statistically confirmed superiority in isokinetic muscle function recovery (Bonferroni p < 0.05 vs. NT, FR, and VFR), suggesting its potential as an effective post-exercise recovery strategy for neuromuscular restoration.

1. Introduction

High-intensity or repetitive exercise induces metabolic stress and neuromuscular fatigue in skeletal muscle [1]. Muscle fatigue is characterized by lactate accumulation, impaired excitation–contraction coupling, and neuromuscular inhibition, collectively diminishing force production capacity [1,2]. These processes are particularly relevant in competitive swimming, where athletes may complete multiple high-intensity events within a single session, requiring rapid inter-event recovery [3]. The present study focuses on acute post-exercise fatigue—characterized by blood lactate accumulation, neuromuscular impairment, and perceived exertion—rather than the delayed inflammatory processes associated with delayed-onset muscle soreness (DOMS), which peaks 24–72 h post-exercise [4].

A diverse range of recovery strategies has been proposed to attenuate muscle fatigue and accelerate physiological recuperation, including cold-water immersion [5], sports massage [6,7], compression garments [8], neuromuscular electrical stimulation [9], and foam rolling [10]. Among these, self-myofascial release (SMR) techniques have attracted considerable attention due to their accessibility and ease of application. Foam rollers, vibration foam rollers, and whole-body vibration exercise (WBVE) have been increasingly adopted as recovery tools in both clinical and athletic settings [10,11,12]. Although cold-water immersion and sports massage have demonstrated efficacy, their field applicability is constrained by the need for specialized equipment or trained practitioners. In contrast, self-administered foam rolling and WBV platforms represent practical, athlete-controlled alternatives that warrant systematic comparison.

Foam rolling effectively induces myofascial relaxation and improves lymphatic circulation and autonomic nervous system regulation through mechanical compression, leading to increased joint range of motion [13,14]. The vibration foam roller builds upon conventional foam rolling by incorporating localized vibratory stimulation, thereby amplifying mechanoreceptor activation and facilitating simultaneous effects on muscle spindle and Golgi tendon organ function, reflex muscle contraction, pain inhibition, and neuromuscular recovery [12,15].

WBVE delivers vibratory stimuli ranging from 15 to 50 Hz, stimulating mechanoreceptors and the neuromuscular system while inducing increased blood flow, enhanced muscle activation, and myofascial relaxation [16,17]. Vibration stimulation is known to elicit the tonic vibration reflex (TVR), which activates the neuromuscular feedback loop and may thereby contribute to the recovery of fatigued muscle [18]. Vibration frequencies of 20–30 Hz are considered particularly effective for improving blood flow and modulating pain, whereas frequencies above 40 Hz are more efficacious for enhancing muscle activation [16,19,20].

Despite the growing adoption of vibration-based recovery strategies, direct comparative research between recovery modalities remains insufficient, and the physiological response differences associated with varying vibration frequencies have not been fully elucidated. Accordingly, the purpose of this study was to comparatively analyze the effects of foam rolling, vibration foam rolling, and WBVE at 12 Hz and 20 Hz on VAS scores, blood lactate concentration, and isokinetic muscle joint function following exercise-induced muscle fatigue, and to provide evidence-based guidance for practical application in athletic and training environments. We hypothesized that active vibration-based recovery interventions (VFR, WBV-12, WBV-20) would produce significantly greater reductions in perceived fatigue, blood lactate, and isokinetic performance deficits compared with passive rest.

2. Materials and Methods

2.1. Participants

A total of eight male amateur swimmers volunteered to participate in this study. All participants were fully informed of the study’s purpose, procedures, potential risks, and expected levels of physical exertion prior to enrollment. Written informed consent was obtained from all participants. This study was approved by the Institutional Review Board of Chungnam National University (IRB approval number: 201906-SB-090-01).

This study employed a within-subjects repeated-measures crossover design, in which each participant completed five experimental sessions at one-week intervals. At each session, participants performed light stretching and warm-up activities, followed by a standardized rest period of at least 10 min, after which baseline measurements were obtained, and the fatigue induction protocol was administered. One of five recovery interventions—no treatment (control), FR, VFR, WBV-12, or WBV-20—was applied in a randomized order, with each intervention separated by one week. VAS scores, blood lactate concentration, and isokinetic muscle joint function were measured at three time points: at rest, five minutes post-fatigue, and immediately post-intervention. The order of recovery interventions was independently randomized for each participant using a computer-generated random sequence (random.org; accessed on 15 March 2019). One-week washout periods were selected based on evidence that blood lactate normalizes within 1–2 h and acute neuromuscular fatigue resolves within 48–72 h post-exercise [18,21]. A post hoc power analysis (G*Power 3.1 (Heinrich-Heine-Universität, Düsseldorf, Germany)) confirmed adequate statistical power (1-β = 0.82) for the primary isokinetic outcome (η²g = 0.226, f = 0.54, α = 0.05). Prior to the first session, participants completed a familiarization session practicing all five interventions and the isokinetic protocol. Participants were instructed to avoid strenuous activity for 48 h before each session and to maintain consistent dietary and hydration habits throughout the study. Participants’ physical characteristics are summarized in Table 1, and the overall experimental design is illustrated in Figure 1.

Table 1. Physical characteristics of participants (mean ± SD).

Variables	Age (Years)	Height (cm)	Weight (kg)	BMI (kg/m2)	Fat (%)	Career (Years)
Subjects (n = 8)	25.1 ± 2.7	175.8 ± 5.8	75.7 ± 8.1	24.5 ± 2.7	14.4 ± 6.5	6.1 ± 3.0

Note: Values are presented as mean ± standard deviation. BMI = body mass index.

Figure 1. Experimental design and protocol. FR = foam roller; VFR = vibration foam roller; WBV = whole-body vibration.

2.2. Exercise-Induced Muscle Fatigue

Knee extensor and flexor muscle fatigue of both lower limbs was induced using an isokinetic dynamometer (Cybex 660 Norm System; Cybex Inc., Medway, MA, USA). Participants were seated with the hips flexed at approximately 90°, and the axis of rotation of the dynamometer was aligned with the lateral femoral epicondyle. The fatigue induction protocol, adapted from Kim [22], consisted of one set comprising 5 repetitions at 60°/s, 10 repetitions at 180°/s, and 30 repetitions at 240°/s, performed for four sets for both limbs. The order of right and left limb testing was independently randomized for each participant to control potential order effects. The complete protocol is presented in Table 2.

Table 2. Muscle fatigue protocol of the isokinetic test.

Item	Angular Velocity (°/s)	Repetitions (REP)	Recovery Time	Sets
Right lower extremity	60	5	1 min 30 s	4
Right lower extremity	180	10
Right lower extremity	240	30
Left lower extremity	60	5
Left lower extremity	180	10
Left lower extremity	240	30

Note: Recovery time (1 min 30 s) was applied between sets. Four sets were performed for both right and left lower extremities.

2.3. Recovery Interventions

(1) No Treatment (control): Following the fatigue induction protocol, participants remained seated in a chair for 15 min without any treatment.

(2) Foam Roller (FR): A foam roller (Hyperice VYPER 2.0; Hyperice, Irvine, CA, USA) was employed. The protocol was adapted from Pearcey et al. [21] and targeted six muscle groups: gastrocnemius, gluteus maximus, iliotibial band, hamstrings, hip adductors, and quadriceps. Each group was rolled from origin to insertion for one minute under body weight, followed by a 15 s rest. Rolling cadence was standardized using a metronome at 50 bpm in 3/4 time, as per Pearcey et al. [21]. All sessions were directly supervised by a trained research assistant.

(3) Vibration Foam Roller (VFR): The same device and protocol as the FR condition (Hyperice VYPER 2.0) were used, with the device’s internal vibration function activated at 45 Hz throughout the session [12,13,23]. We note that the Hyperice VYPER 2.0 operates at a single fixed internal frequency of 45 Hz and cannot be adjusted by the user, which differs from the WBV frequencies tested (12 and 20 Hz).

(4) Whole-Body Vibration at 12 Hz (WBV-12): A vibratory platform (Galileo; Novotec Medical GmbH, Pforzheim, Germany) delivered vibration at 12 Hz with an amplitude of 4 mm. Each bout consisted of 2 min of vibration followed by 1 min of rest, for five sets (15 min total). Participants maintained an upright torso posture with the knee joint flexed to approximately 120° [18,24].

(5) Whole-Body Vibration at 20 Hz (WBV-20): The same vibratory platform was used with the frequency adjusted to 20 Hz at an amplitude of 4 mm. The session structure was identical to WBV-12 (five sets of 2 min vibration and 1 min rest; 15 min total), maintaining the same standardized posture [18].

2.4. Outcome Measures

Perceived Fatigue Assessment: Perceived fatigue was assessed using a 100 mm visual analog scale (VAS). Participants were asked: “Please mark the level of fatigue you are currently experiencing.” The left anchor was labeled “No fatigue whatsoever” (0 mm) and the right anchor “Extreme fatigue” (100 mm). The distance from the left anchor was measured to the nearest millimeter; all VAS values in this paper are reported in mm. Assessments were conducted at three time points: at rest, five minutes post-fatigue, and immediately post-intervention.

Blood Lactate Concentration: Blood samples were collected from fingertip capillaries using a standardized single-use lancet. Approximately 25 μL of capillary blood was collected and immediately analyzed using a portable lactate analyzer (Biosen C-line; EKF Diagnostics, Barleben, Germany; measurement range: 0.5–40 mmol/L; CV < 3%). All samples were analyzed by the same investigator to minimize inter-operator variability.

Isokinetic Muscle Function: Isokinetic peak torque was measured using the Cybex 660 Norm System. The range of motion was restricted to 0° to 100°. Participants were seated with the knee joint axis aligned with the dynamometer’s axis of rotation. An isokinetic angular velocity of 60°/s was selected, with five repetitions per set. Knee extensor and flexor peak torque (%BW) was assessed after the final fatigue set and after each recovery intervention. Right- and left-limb peak torque values were averaged to yield a single bilateral composite score. Pre-fatigue baseline isokinetic measurements were obtained prior to the first experimental session to confirm fatigue induction efficacy.

2.5. Statistical Analysis

All data were analyzed using SPSS Statistics version 24.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics are reported as mean ± standard deviation (SD). Data are additionally presented with 95% confidence intervals (95% CI). A two-way repeated-measures ANOVA (intervention × time) was applied to VAS and blood lactate data to assess the main effects of recovery methods at each time point. For isokinetic outcomes, one-way repeated-measures ANOVAs were conducted separately at the post-fatigue and post-recovery time points to verify pre-intervention condition equivalence and to assess post-intervention recovery effects; baseline-to-post-fatigue differences were assessed using paired t-tests. Where Mauchly’s test indicated a violation of sphericity, the Greenhouse–Geisser correction was applied. Post hoc pairwise comparisons were conducted using the Bonferroni correction to control for family-wise error rate. Statistical significance was set at α = 0.05. Effect sizes were calculated using generalized eta squared (η²g) and Cohen’s d and interpreted according to conventional thresholds: small ≥ 0.01, medium ≥ 0.06, and large ≥ 0.14.

3. Results

3.1. Fatigue Protocol Verification

The isokinetic fatigue protocol produced a significant bilateral reduction in extensor peak torque from baseline (229.2 ± 37.8%BW) to the post-fatigue measurement (189.8 ± 27.5%BW; t(7) = 4.19, p = 0.004, d = 1.48), confirming successful fatigue induction prior to the recovery interventions.

3.2. Changes in VAS Scores After Recovery Interventions

VAS scores at each time point are presented in Table 3 and Figure 2. No significant main effect was observed at rest (F(4,28) = 1.04, p = 0.407, η²g = 0.056) or five minutes post-fatigue (F(4,28) = 0.83, p = 0.517, η²g = 0.016), confirming comparable baseline equivalence and fatigue levels. A significant main effect was observed following recovery interventions (F(4,28) = 5.98, p = 0.001, η²g = 0.248). Post-recovery VAS scores (mm): NT = 57.5 ± 8.9, FR = 38.8 ± 18.1, VFR = 33.8 ± 14.1, WBV-12 = 38.8 ± 17.3, and WBV-20 = 38.8 ± 16.4. All active conditions showed substantially lower perceived fatigue than NT (Cohen’s d = 0.82–1.23, large effects); however, individual Bonferroni pairwise comparisons did not reach adjusted significance, reflecting limited statistical power with n = 8.

Table 3. Effects of recovery interventions on perceived fatigue (mean ± SD).

Time Point	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
Rest	13.8 ± 10.6	16.2 ± 9.2	17.5 ± 13.9	21.2 ± 15.5	21.2 ± 13.6	1.035	0.407	0.056	ns
PF	72.5 ± 11.6	75.0 ± 14.1	70.0 ± 16.0	72.5 ± 15.8	75.0 ± 19.3	0.831	0.517	0.016	ns
PR	57.5 ± 8.9	38.8 ± 18.1	33.8 ± 14.1	38.8 ± 17.3	38.8 ± 16.4	5.984	0.001 **	0.248	ns (all pairs) ‡

Note: Values are mean ± SD (mm). η²g = generalized eta squared. Statistical analysis: two-way repeated-measures ANOVA (intervention × time) with Bonferroni-adjusted post hoc pairwise comparisons across the five recovery conditions (k = 10 pairs per time point). Symbols used in this table: ** omnibus F-test p < 0.01; ns = non-significant Bonferroni-adjusted pairwise comparison (p > 0.05). ‡ “ns (all pairs)” at post-recovery explicitly indicates that although the omnibus F-test was significant (F(4,28) = 5.98, p = 0.001, η²g = 0.248), none of the 10 individual Bonferroni-adjusted pairwise comparisons reached p < 0.05; all active conditions nonetheless showed large effect sizes versus NT (Cohen’s d = 0.82–1.23), indicating clinically meaningful but statistically unconfirmed differences at the current sample size (n = 8). Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (whole-body vibration at 12 Hz); e = WBV-20 (whole-body vibration at 20 Hz). PF = post-fatigue; PR = post-recovery.

Figure 2. Changes in perceived fatigue (VAS, mm) across rest, post-fatigue (baseline = at rest), and post-recovery for five recovery conditions. Error bars = SD. * Significant omnibus main effect at post-recovery: F(4,28) = 5.98, p = 0.001, η²g = 0.248. Note: Visual differences between conditions at post-recovery reflect large effect sizes (d = 0.82–1.23) but do not indicate statistically confirmed pairwise differences; no individual Bonferroni comparisons reached adjusted significance. NT = no treatment (control); FR = foam roller; VFR = vibration foam roller; WBV-12 = WBV 12 Hz; WBV-20 = WBV 20 Hz.

3.3. Changes in Blood Lactate Concentration After Recovery Interventions

Blood lactate concentrations at each time point are presented in Table 4 and Figure 3. No significant main effect was observed at rest (F(4,28) = 0.69, p = 0.604, η²g = 0.046) or five minutes post-fatigue (F(4,28) = 0.18, p = 0.948, η²g = 0.007). A significant main effect was observed following recovery interventions (F(4,28) = 5.12, p = 0.003, η²g = 0.226). Post-recovery concentrations (mmol/L, 95% CI): NT = 5.72 ± 1.79 [4.22, 7.22], FR = 4.70 ± 1.47 [3.47, 5.93], VFR = 3.84 ± 1.28 [2.77, 4.91], WBV-12 = 5.16 ± 1.51 [3.89, 6.42], and WBV-20 = 3.92 ± 0.95 [3.12, 4.71]. VFR and WBV-20 showed the largest reductions vs NT (Cohen’s d = 1.01 and 1.13); individual Bonferroni pairwise comparisons did not reach adjusted significance.

Figure 3. Changes in blood lactate concentration (mmol/L) across rest, post-fatigue (baseline = at rest), and post-recovery for five recovery conditions. Error bars = SD. * Significant omnibus main effect at post-recovery: F(4,28) = 5.12, p = 0.003, and η²g = 0.226. Note: Visual differences between conditions at post-recovery do not indicate statistically confirmed pairwise differences; no individual Bonferroni comparisons for raw lactate reached adjusted significance (see Table 5 for confirmed lactate recovery rate comparisons). NT = no treatment; FR = foam roller; VFR = vibration foam roller; WBV-12 = WBV 12 Hz; WBV-20 = WBV 20 Hz.

Figure 3. Changes in blood lactate concentration (mmol/L) across rest, post-fatigue (baseline = at rest), and post-recovery for five recovery conditions. Error bars = SD. * Significant omnibus main effect at post-recovery: F(4,28) = 5.12, p = 0.003, and η²g = 0.226. Note: Visual differences between conditions at post-recovery do not indicate statistically confirmed pairwise differences; no individual Bonferroni comparisons for raw lactate reached adjusted significance (see Table 5 for confirmed lactate recovery rate comparisons). NT = no treatment; FR = foam roller; VFR = vibration foam roller; WBV-12 = WBV 12 Hz; WBV-20 = WBV 20 Hz.

Table 4. Comparison of blood lactate concentration across recovery methods (mean ± SD, mmol/L).

Time Point	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
Rest	1.74 ± 0.44	1.80 ± 1.26	1.59 ± 0.36	1.47 ± 0.43	1.44 ± 0.57	0.69	0.604	0.046	ns
PF	10.15 ± 1.95	10.00 ± 1.88	10.07 ± 2.37	10.41 ± 2.18	9.95 ± 1.36	0.18	0.948	0.007	ns
PR	5.72 ± 1.79	4.70 ± 1.47	3.84 ± 1.28	5.16 ± 1.51	3.92 ± 0.95	5.12	0.003 *	0.226	ns (all pairs) ‡

Note: Values are mean ± SD (mmol/L). η²g = generalized eta squared. Statistical analysis: two-way repeated-measures ANOVA (intervention × time) with Bonferroni-adjusted post hoc pairwise comparisons across the five recovery conditions (k = 10 pairs per time point). Symbols used in this table: * omnibus F-test p < 0.05; ‡ “ns (all pairs)” at post-recovery explicitly indicates that although the omnibus F-test was significant (F(4,28) = 5.12, p = 0.003, η²g = 0.226), none of the 10 individual Bonferroni-adjusted pairwise comparisons for raw lactate concentration reached p < 0.05; VFR and WBV-20 nonetheless showed large effect sizes versus NT (Cohen’s d = 1.01 and 1.13, respectively), indicating clinically meaningful but statistically unconfirmed differences at the current sample size (n = 8). Statistically confirmed pairwise differences were observed for the derived lactate recovery rate variable (see Table 5). Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (12 Hz); e = WBV-20 (20 Hz). PF = post-fatigue; PR = post-recovery.

Table 4. Comparison of blood lactate concentration across recovery methods (mean ± SD, mmol/L).

Time Point	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
Rest	1.74 ± 0.44	1.80 ± 1.26	1.59 ± 0.36	1.47 ± 0.43	1.44 ± 0.57	0.69	0.604	0.046	ns
PF	10.15 ± 1.95	10.00 ± 1.88	10.07 ± 2.37	10.41 ± 2.18	9.95 ± 1.36	0.18	0.948	0.007	ns
PR	5.72 ± 1.79	4.70 ± 1.47	3.84 ± 1.28	5.16 ± 1.51	3.92 ± 0.95	5.12	0.003 *	0.226	ns (all pairs) ‡

Note: Values are mean ± SD (mmol/L). η²g = generalized eta squared. Statistical analysis: two-way repeated-measures ANOVA (intervention × time) with Bonferroni-adjusted post hoc pairwise comparisons across the five recovery conditions (k = 10 pairs per time point). Symbols used in this table: * omnibus F-test p < 0.05; ‡ “ns (all pairs)” at post-recovery explicitly indicates that although the omnibus F-test was significant (F(4,28) = 5.12, p = 0.003, η²g = 0.226), none of the 10 individual Bonferroni-adjusted pairwise comparisons for raw lactate concentration reached p < 0.05; VFR and WBV-20 nonetheless showed large effect sizes versus NT (Cohen’s d = 1.01 and 1.13, respectively), indicating clinically meaningful but statistically unconfirmed differences at the current sample size (n = 8). Statistically confirmed pairwise differences were observed for the derived lactate recovery rate variable (see Table 5). Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (12 Hz); e = WBV-20 (20 Hz). PF = post-fatigue; PR = post-recovery.

Table 5. Comparison of blood lactate recovery rate across recovery methods (mean ± SD, %).

Item	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
PR	52.18 ± 16.82	67.45 ± 15.30	75.10 ± 8.76	59.73 ± 8.02	70.82 ± 8.02	5.37	0.003 **	0.346	a < c (p = 0.031) *; a < e (p = 0.027) *

Note: Values are mean ± SD (%). η²g = generalized eta squared. Statistical analysis: one-way repeated-measures ANOVA with Bonferroni-adjusted post hoc pairwise comparisons restricted to pre-specified contrasts versus NT (k = 4; not all-pairwise). Symbols used in this table: ** omnibus F-test p < 0.01; * Bonferroni-adjusted pairwise p < 0.05 versus NT. Exact Bonferroni-adjusted p-values for significant contrasts: VFR versus NT, p = 0.031 (Cohen’s d = 1.01); WBV-20 versus NT, p = 0.027 (Cohen’s d = 1.13). FR versus NT (p > 0.05) and WBV-12 versus NT (p > 0.05) did not reach Bonferroni-adjusted significance. Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (12 Hz); e = WBV-20 (20 Hz). PR = post-recovery.

Table 5. Comparison of blood lactate recovery rate across recovery methods (mean ± SD, %).

Item	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
PR	52.18 ± 16.82	67.45 ± 15.30	75.10 ± 8.76	59.73 ± 8.02	70.82 ± 8.02	5.37	0.003 **	0.346	a < c (p = 0.031) *; a < e (p = 0.027) *

Note: Values are mean ± SD (%). η²g = generalized eta squared. Statistical analysis: one-way repeated-measures ANOVA with Bonferroni-adjusted post hoc pairwise comparisons restricted to pre-specified contrasts versus NT (k = 4; not all-pairwise). Symbols used in this table: ** omnibus F-test p < 0.01; * Bonferroni-adjusted pairwise p < 0.05 versus NT. Exact Bonferroni-adjusted p-values for significant contrasts: VFR versus NT, p = 0.031 (Cohen’s d = 1.01); WBV-20 versus NT, p = 0.027 (Cohen’s d = 1.13). FR versus NT (p > 0.05) and WBV-12 versus NT (p > 0.05) did not reach Bonferroni-adjusted significance. Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (12 Hz); e = WBV-20 (20 Hz). PR = post-recovery.

3.4. Changes in Blood Lactate Recovery Rate After Recovery Interventions

The blood lactate recovery rate was calculated as follows: recovery rate (%) = [(lactate post-fatigue–lactate post-intervention)/(lactate post-fatigue–lactate at rest)] × 100. Results are presented in Table 5. A significant main effect was observed (F(4,28) = 5.37, p = 0.003, η²g = 0.346). Post hoc Bonferroni analysis (k = 4 comparisons vs. NT) indicated that the lactate recovery rate was significantly higher in VFR (75.10 ± 8.76%, p = 0.031) and WBV-20 (70.82 ± 8.02%, p = 0.027) conditions compared to NT (52.18 ± 16.82%). FR (67.45 ± 15.30%) and WBV-12 (59.72 ± 8.02%) showed higher rates but did not reach Bonferroni-adjusted significance.

3.5. Changes in Isokinetic Muscle Function After Recovery Interventions

Bilateral-averaged (right + left) isokinetic extensor peak torque data are presented in Table 6 and Figure 4. Pre-fatigue baseline extensor peak torque was 229.2 ± 37.8%BW (common to all conditions). At the post-fatigue time point, a significant main effect of condition was observed (F(4,28) = 4.87, p = 0.004, η²g = 0.091), with Bonferroni analysis indicating that NT and VFR showed lower torque than WBV-12 (a,c < d; p < 0.05). A one-way repeated-measures ANOVA at the post-recovery time point revealed a significant main effect (F(4,28) = 10.75, p < 0.001, η²g = 0.226). Post-recovery bilateral extensor values (%BW): NT = 180.2 ± 32.7, FR = 193.2 ± 34.6, VFR = 214.1 ± 43.5, WBV-12 = 222.8 ± 48.9, and WBV-20 = 239.9 ± 47.6. WBV-20 was the only condition to exceed the pre-fatigue baseline level at post-recovery (239.9 vs. 229.2%BW). Bonferroni post hoc indicated that WBV-20 produced significantly higher peak torque than NT (p = 0.031, d = 1.56), FR (p = 0.019, d = 1.71), and VFR (p = 0.017), i.e., a,b,c < e (p < 0.05).

Table 6. Bilateral averaged comparison of 60°/s peak torque (%BW) of knee extensor across recovery methods (mean ± SD).

Item	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
Baseline	229.2 ± 37.8	229.2 ± 37.8	229.2 ± 37.8	229.2 ± 37.8	229.2 ± 37.8	—	—	—	ns
PF	189.8 ± 27.5 a	187.1 ± 28.9 ab	193.7 ± 24.3 a	203.8 ± 23.2 b	207.7 ± 29.9 ab	4.87	0.004 **	0.091	a, c < d (Bonferroni p < 0.05) *
PR	180.2 ± 32.7 a	193.2 ± 34.6 a	214.1 ± 43.5 a	222.8 ± 48.9 ab	239.9 ± 47.6 b	10.75	<0.001 **	0.226	a, b, c < e (p = 0.031, 0.019, 0.017) *

Note: Values are mean ± SD (%BW). η²g = generalized eta squared. Statistical analysis: one-way repeated-measures ANOVA applied separately at post-fatigue and post-recovery, with Bonferroni-adjusted all-pairwise post hoc comparisons across the five recovery conditions (k = 10 pairs per time point). Symbols used in this table: ** omnibus F-test p < 0.01; * Bonferroni-adjusted pairwise p < 0.05. Superscript letters indicate Bonferroni-adjusted all-pairwise differences, within a row: conditions not sharing a common superscript letter differ at Bonferroni-adjusted p < 0.05; conditions sharing a common letter are not statistically distinguishable. At post-recovery, WBV-20 (e) produced significantly higher extensor peak torque than NT (a, p = 0.031, Cohen’s d = 1.56), FR (b, p = 0.019, d = 1.71), and VFR (c, p = 0.017). At post-fatigue, WBV-12 (d) showed higher extensor peak torque than NT (a) and VFR (c) (Bonferroni p < 0.05), reflecting residual day-to-day neuromuscular variability despite randomization; an ANCOVA using post-fatigue as a covariate is discussed in the Limitations section. Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (whole-body vibration at 12 Hz); e = WBV-20 (whole-body vibration at 20 Hz). PF = post-fatigue; PR = post-recovery.

Figure 4. Changes in bilateral-averaged knee extensor peak torque (%BW) at baseline, post-fatigue (fourth set), and post-recovery for five recovery conditions. Error bars = SD. ** Significant main effect at post-recovery: F(4,28) = 10.75, p < 0.001, and η²g = 0.226. WBV-20 is significantly higher than NT (Bonferroni-adjusted p = 0.031), FR (p = 0.019), and VFR (p = 0.017). NT = no treatment; FR = foam roller; VFR = vibration foam roller; WBV-12 = WBV 12 Hz; WBV-20 = WBV 20 Hz.

Bilateral-averaged knee flexor peak torque data are presented in Table 7. Pre-fatigue baseline flexor peak torque was 124.4 ± 27.3%BW (common to all conditions). No significant main effect was observed at post-fatigue (F(4,28) = 1.025, p = 0.411, η²g = 0.039). A significant main effect was observed at post-recovery (F(4,28) = 5.082, p = 0.003, η²g = 0.170). Post-recovery bilateral flexor values (%BW): NT = 96.4 ± 25.3, FR = 97.4 ± 21.5, VFR = 110.7 ± 23.5, WBV-12 = 114.6 ± 28.1, and WBV-20 = 124.8 ± 27.9. Notably, WBV-20 was the only condition to fully recover to baseline level (124.8 vs. 124.4%BW). Bonferroni post hoc indicated that FR showed significantly lower flexor torque than WBV-20 (b < e; p < 0.05, d = 1.31).

Table 7. Bilateral averaged comparison of 60°/s peak torque (%BW) of knee flexor across recovery methods (mean ± SD).

Item	a (Control)	b (FR)	c (VFR)	d (WBV-12)	e (WBV-20)	F	p	η2g	Post Hoc
Baseline	124.4 ± 27.3	124.4 ± 27.3	124.4 ± 27.3	124.4 ± 27.3	124.4 ± 27.3	—	—	—	ns
PF	100.2 ± 25.6	98.4 ± 24.7	105.6 ± 17.5	104.9 ± 21.1	109.9 ± 18.8	1.025	0.411	0.039	ns
PR	96.4 ± 25.3 ab	97.4 ± 21.5 a	110.7 ± 23.5 ab	114.6 ± 28.1 ab	124.8 ± 27.9 b	5.082	0.003 **	0.170	b < e (Bonferroni p < 0.05, d = 1.31) *

Note: Values are mean ± SD (%BW). η²g = generalized eta squared. Statistical analysis: one-way repeated-measures ANOVA applied separately at post-fatigue and post-recovery, with Bonferroni-adjusted all-pairwise post hoc comparisons across the five recovery conditions (k = 10 pairs per time point). Symbols used in this table: ** omnibus F-test p < 0.01; * Bonferroni-adjusted pairwise p < 0.05; ns = non-significant Bonferroni-adjusted pairwise comparison (p > 0.05). Superscript letters indicate Bonferroni-adjusted all-pairwise differences, within a row: conditions not sharing a common superscript letter differ at Bonferroni-adjusted p < 0.05; conditions sharing a common letter are not statistically distinguishable. At post-recovery, WBV-20 (e) produced significantly higher flexor peak torque than FR (b) (Bonferroni p < 0.05, Cohen’s d = 1.31). Column labels: a = NT (no treatment/control); b = FR (foam roller); c = VFR (vibration foam roller); d = WBV-12 (whole-body vibration at 12 Hz); e = WBV-20 (whole-body vibration at 20 Hz). PF = post-fatigue; PR = post-recovery.

4. Discussion

The principal findings of this study were: (1) All active recovery interventions produced lower perceived fatigue scores than passive rest at the post-recovery time point (F(4,28) = 5.984, p = 0.001, η²g = 0.248) with large effect sizes (d = 0.82–1.23 vs. NT), although no individual Bonferroni pairwise comparisons reached adjusted significance—these VAS findings are therefore regarded as preliminary. (2) VFR and WBV-20 produced significantly higher lactate recovery rates than NT (Bonferroni p < 0.05), while raw blood lactate pairwise comparisons did not reach significance. (3) WBV-20 produced the greatest recovery of bilateral isokinetic extensor peak torque (F(4,28) = 10.75, p < 0.001, η²g = 0.226), achieving values significantly higher than NT, FR, and VFR at post-recovery (Bonferroni p < 0.05). The isokinetic torque outcomes represent the most statistically robust evidence in this study. Notably, WBV-20 was the only condition in which post-recovery extensor peak torque (239.9 ± 47.6%BW) exceeded the pre-fatigue baseline value (229.2 ± 37.8%BW), suggesting a possible supercompensatory neuromuscular response. To the authors’ knowledge, this is the first study to directly compare foam rolling, vibration foam rolling, and two frequencies of whole-body vibration in a single crossover design in amateur swimmers.

It is important to clarify that the present interventions targeted acute post-exercise fatigue rather than delayed-onset muscle soreness (DOMS). DOMS peaks 24–72 h post-exercise and was not the target of these immediate 15 min interventions. VAS scores reflect perceived acute exertion rather than delayed soreness. Future studies should include follow-up assessments at 24 and 48 h to examine longer-term recovery trajectories.

Amateur swimmers were selected because competitive swimming involves repeated high-intensity events within a single meet, where rapid neuromuscular and metabolic recovery may directly influence performance. Lower-extremity muscular fatigue is a recognized limiting factor in swimming, as the kicking action contributes substantially to propulsive force. These findings therefore have direct practical relevance to athletes and coaches.

Recovery from acute exercise-induced muscle fatigue encompasses restoration of neuromuscular stability, maintenance of metabolic homeostasis, and normalization of muscle functional capacity. The present study compared five recovery interventions and demonstrated that vibration-based modalities—specifically VFR and WBV-20—produced significant improvements across multiple physiological recovery indices, including VAS scores, blood lactate concentration, and isokinetic muscle joint function. These findings may be attributed to the distinct yet complementary physiological mechanisms of foam rolling and vibratory stimulation, including mechanical compression, mechanoreceptor stimulation, muscle spindle activation, and autonomic nervous system modulation.

With respect to VAS scores, a significant main effect of intervention was observed at the post-recovery time point (F(4,28) = 5.98, p = 0.001, η²g = 0.248); however, individual Bonferroni-adjusted pairwise comparisons did not reach statistical significance, and these findings should therefore be regarded as preliminary. All active conditions (FR, VFR, WBV-12, and WBV-20) showed notably lower perceived fatigue than NT with large effect sizes (Cohen’s d = 0.82–1.23), suggesting clinically meaningful differences that may have lacked statistical power given the small sample (n = 8). These trends are consistent with prior research indicating that foam rolling and vibration-based strategies may reduce perceived fatigue [4,13]. Mechanistically, foam rolling is thought to reduce myofascial adhesions through localized mechanical compression and to activate the parasympathetic nervous system to promote autonomic stabilization [13,25]. Vibratory stimulation may act upon peripheral mechanoreceptors—particularly Aβ mechanoreceptors and proprioceptors within muscle spindles—to modulate pain transmission at the spinal dorsal horn [3]. Vibration may also reduce the excitability of Aδ and C fiber nociceptive pathways and activate inhibitory interneurons at the spinal level, contributing to gate control analgesia [26,27]. Confirmatory evidence from larger samples is needed before conclusions about pairwise superiority can be drawn.

With respect to blood lactate concentration, a significant main effect of intervention was observed at post-recovery (F(4,28) = 5.12, p = 0.003, η²g = 0.226); however, as with VAS, individual Bonferroni pairwise comparisons for raw lactate concentrations did not reach adjusted significance. Statistically confirmed differences were observed only for the derived lactate recovery rate variable, where VFR (p = 0.031) and WBV-20 (p = 0.027) produced significantly higher recovery rates than NT. FR and WBV-12 showed numerically higher rates but did not reach Bonferroni-adjusted significance. These blood lactate findings should therefore be interpreted as preliminary, with large effect sizes (d = 1.01–1.13 for VFR and WBV-20 vs. NT) indicating potential clinical relevance that warrants confirmation in larger samples. Vibration-based interventions are hypothesized to enhance local blood flow in skin and fascial tissues and to stimulate lymphatic circulation, thereby accelerating the clearance of accumulated lactate and hydrogen ions (H⁺) from muscle fibers [5]. The superior lactate recovery observed in WBV-20 compared with WBV-12 may be attributable to frequency-dependent differences in circulatory and neuromuscular responses; at 20 Hz, vibration stimuli are more closely aligned with the resonant frequency of human lower-limb musculature, promoting greater muscle spindle activation, increased capillary recruitment, and more effective intramuscular blood perfusion [16,19]. In contrast, 12 Hz may be insufficient to elicit the full tonic vibration reflex (TVR) response required to sustain elevated blood flow and metabolic clearance during the recovery period [18]. These mechanistic differences provide a plausible physiological basis for the differential lactate recovery observed between WBV-20 and WBV-12, although direct comparisons of circulatory responses at these two frequencies in healthy athletes remain to be established.

Regarding muscle functional recovery, vibration-based interventions yielded superior outcomes across multiple neuromuscular indices, with WBV-20 demonstrating statistically confirmed superiority over NT, FR, and VFR for bilateral extensor peak torque at post-recovery (Bonferroni p < 0.05). These findings align with prior studies [18,28,29] demonstrating that vibration stimulation activates group Ia afferent fibers within muscle spindles, enhancing neuromuscular function. By increasing α-motor neuron responsiveness and eliciting the TVR, vibratory stimulation augments neuromuscular excitability [18]. The TVR is particularly important during fatigued states, as it transiently restores the voluntary drive of the central nervous system, re-facilitating voluntary muscle contractions [30]. Furthermore, vibration accelerates motor unit recruitment and increases the discharge rate of alpha motor neurons, positively influencing fast-twitch muscle fiber activation [29]. These isokinetic torque outcomes provide the most statistically robust evidence in the present study and support the conclusion that WBV-20 is superior to passive rest and foam rolling-based strategies for acute neuromuscular recovery.

The consistent superiority of WBV-20 over WBV-12 across multiple outcome variables warrants mechanistic discussion. Vibration frequency is a critical determinant of the neuromuscular and circulatory responses to WBV [16,19]. At 20 Hz, the mechanical oscillation frequency falls within the range that most effectively entrains the tonic vibration reflex (TVR) via Ia afferent spindle activation, thereby sustaining α-motor neuron excitability throughout the recovery period [18]. In contrast, WBV at 12 Hz may generate a sub-optimal stimulus for TVR induction, resulting in attenuated neuromuscular re-facilitation. From a circulatory perspective, 20 Hz stimulation has been shown to increase femoral artery blood flow velocity more effectively than lower frequencies, promoting greater capillary recruitment and metabolic waste clearance [16]. Additionally, at 20 Hz, muscle spindle afferents are activated at a rate more aligned with natural physiological motor unit discharge patterns (10–30 Hz), facilitating synchronized motor unit recruitment and reducing the energetic cost of re-establishing neuromuscular force production [19,31]. These frequency-dependent differences collectively explain why WBV-20 produced greater recovery of both isokinetic torque and lactate clearance rates compared to WBV-12 in the present study, underscoring the importance of frequency optimization when applying WBV as a post-exercise recovery tool.

Several limitations should be acknowledged. The sample was restricted to young male amateur swimmers (n = 8), which substantially limits statistical power for individual pairwise comparisons and the generalizability of the findings; results should be regarded as preliminary evidence requiring replication with larger, more diverse samples including female athletes. The absence of significant Bonferroni pairwise comparisons for VAS and blood lactate, despite a significant omnibus F, reflects the conservative nature of Bonferroni correction with k = 10 pairs and n = 8; large observed effect sizes (d = 0.82–1.23) suggest clinically meaningful differences that would likely reach significance with larger samples. Future research should incorporate participants of diverse sexes, ages, and training backgrounds, and should integrate multi-layered physiological assessments, including inflammatory markers, oxidative stress indicators, and hormonal responses, to provide a more comprehensive evaluation of these recovery modalities. Additionally, although the within-subjects crossover design with a randomized condition order was employed to control session-to-session variability, a significant difference among conditions at the post-fatigue time point was observed for isokinetic extensor peak torque (F(4,28) = 4.87, p = 0.004), likely reflecting natural day-to-day fluctuations in neuromuscular output; therefore, future studies should consider covariate-adjusted analyses (ANCOVA) using post-fatigue values to more rigorously isolate the intervention effect.

5. Conclusions

The present study provides preliminary evidence suggesting that among the various post-exercise recovery interventions examined, the evidence for superiority was heterogeneous and outcome-dependent. For isokinetic muscle function—the outcome with the strongest statistical support—WBV-20 produced significantly greater recovery of bilateral extensor peak torque than NT, FR, and VFR (Bonferroni p < 0.05), and was the only condition to exceed the pre-fatigue baseline at post-recovery. For blood lactate recovery rate, VFR and WBV-20 produced significantly higher recovery rates than NT (Bonferroni p < 0.05). For VAS scores and raw blood lactate concentrations, omnibus ANOVA effects were significant, but individual Bonferroni pairwise comparisons did not reach adjusted significance; these findings should be regarded as exploratory, supported by large effect sizes (d = 0.82–1.23) that suggest clinical relevance but require confirmation in larger samples.

WBV-20 demonstrated the most consistent and statistically supported recovery effects across the dependent variables examined, suggesting that it may simultaneously facilitate neuromuscular re-facilitation via TVR-mediated α-motor neuron activation and metabolic recovery via enhanced circulatory support. The frequency-dependent advantage of WBV-20 over WBV-12 observed across both metabolic and neuromuscular outcomes highlights the importance of vibration parameter optimization in applied recovery settings. Compared with passive rest, WBV-20 may represent a strategically advantageous post-exercise recovery modality in athletic contexts requiring rapid inter-event neuromuscular restoration. However, these conclusions should be interpreted with caution given the small sample size (n = 8); further research with larger samples is warranted.