Height, Sex, and Sport as Correlates of Tendon Stiffness in Elite Athletes

Abstract

Background/Objectives: Understanding the factors that influence tendon mechanical properties is essential for optimizing performance and preventing injuries in elite athletes. This study aimed to identify the strongest correlates of the biomechanical properties (frequency, stiffness, logarithmic decrement, relaxation, and creep) in the Achilles and patellar tendons in elite international athletes. Methods: A cross-sectional study was conducted with 111 elite athletes from 11 sports disciplines assessed at a high-performance training center. Tendon properties were measured bilaterally using MyotonPRO. Anthropometric (height, weight, age), demographic (sex, limb dominance defined as the preferred limb for sport-specific activities), and sport-specific variables were analyzed using correlation, multiple regression, and machine learning approaches. Results: Height showed the strongest correlations with tendon frequency and stiffness, particularly for the Achilles tendon (r = 0.52 for frequency; r = 0.53 for stiffness; p ≤ 0.001, large effects). Sex differences were evident across all measures, with men showing higher stiffness and frequency, and women greater relaxation and creep (partial η² = 0.35–0.48, Cohen’s d = 0.84–1.16). Sports discipline explained substantial variance in tendon properties (η² > 0.40), and limb dominance influenced Achilles stiffness, with left-dominant athletes showing higher values (p < 0.05). Age showed minimal associations (r < 0.10). Conclusions: Height, sex, and sports discipline were the strongest correlates of Achilles and patellar tendon mechanical properties in elite athletes, with large and practically meaningful effects across sports. This comprehensive analysis, utilizing multivariate and machine learning approaches, provides insights that can inform individualized training, injury prevention, and performance optimization strategies in high-performance sports.

1. Introduction

Tendon mechanical adaptations vary with sport, sex, and anthropometry, yet comparative data across elite disciplines remain scarce [1,2,3]. The Achilles and patellar tendons serve distinct functions: the Achilles primarily stores and releases elastic energy during running and jumping, while the patellar tendon transmits quadriceps forces during knee extension [3,4]. These functional differences lead to sport- and sex-specific adaptations, with males generally exhibiting stiffer tendons and greater adaptation variability across sports [1,2,3,5,6]. Tendon biomechanical properties critically influence athletic performance and injury risk [7,8,9]. The Myoton PRO (Myoton AS, Estonia) device provides non-invasive assessment of five complementary parameters: frequency (natural oscillation), stiffness (resistance to deformation), logarithmic decrement (vibration decay), relaxation (tension reduction), and creep (progressive deformation under load) [7,10,11,12]. While ultrasound elastography remains the gold standard for tendon assessment, Myoton offers distinct advantages for elite athlete populations: (1) field-portable assessment without imaging equipment, (2) rapid bilateral evaluation (<5 min vs. 20–30 min), (3) operator-independent measurements reducing inter-examiner variability, and (4) real-time feedback capability for training environments. Validation studies demonstrate excellent reliability (ICC 0.74–0.99) and strong correlations with elastography (r = 0.72–0.89), supporting its use as a practical alternative for high-performance sports contexts [8,13,14,15].

The use of this equipment is interesting in high-performance sports, as tendon stiffness plays a central role in athletic performance by optimizing the storage and release of elastic energy during movement [10]. In this context, understanding the correlation of tendon biomechanical properties is critically important because of the extreme mechanical demands imposed by high-intensity training and competition [16,17]. However, stiffness is a double-edged characteristic; while adequate stiffness enhances power transmission and performance, excessive stiffness may compromise shock absorption and increase injury risk [18,19,20]. Thus, optimal tendon behavior represents a balance between performance efficiency and mechanical protection [21]. Sport-specific adaptations illustrate this principle: sprint and power athletes generally display higher tendon stiffness to maximize explosive performance, whereas endurance athletes present more compliant tendon profiles, which favor energy economy over prolonged periods [22,23]. Despite these general trends, the extent to which different sports, sex, and anthropometric characteristics interact to shape tendon mechanics in elite populations remains poorly defined.

Deviations from sport-specific tendon mechanical optima increase tendinopathy risk, particularly in the frequently injured Achilles and patellar tendons [9,24,25]. Key correlates include sex (males exhibit greater stiffness, females greater compliance due to hormonal and morphological factors [26,27], anthropometry (taller athletes develop stiffer profiles under greater mechanical demands [28] and age (though elite training may modify typical age-related stiffening patterns [29]. Another understudied correlation is limb dominance, which may lead to asymmetries in tendon properties due to sport-specific unilateral loading [30,31]. However, few studies have systematically examined the laterality effects in elite populations across different sports disciplines. A recent study with highly trained athletes demonstrated that a 12-week heavy resistance training program increased Achilles and patellar tendon stiffness by 39% and 16%, respectively, alongside elevated biochemical markers of extracellular matrix remodeling [32]. These findings indicate that high-intensity exercise can promote adaptive tendon stiffening, which may counteract the age-related loss of compliance typically observed in sedentary individuals [22,29]. Nevertheless, the extent to which chronic, sport-specific loading in elite athletes offsets or modifies age-related changes remains unclear, given that their tendons are already exposed to extreme mechanical demands and long-term adaptations.

Current literature limitations include the scarcity of multi-sport elite athlete data and the lack of comprehensive multivariate analyses of tendon mechanical correlates [33], highlighting the need for systematic investigation across diverse elite sporting populations. While previous Myoton studies in elite athletes have focused on single sports [4,34,35] or limited parameter sets, no study has comprehensively examined all five Myoton parameters across multiple elite sports with a systematic analysis of anthropometric and demographic correlates. Therefore, this study aimed to: (1) provide the first comprehensive multivariate analysis of Myoton-derived tendon properties across diverse elite sporting disciplines, (2) establish sport-specific and anthropometric correlates using advanced statistical modeling, including machine learning approaches, and (3) differentiate mechanical property patterns between Achilles and patellar tendons in elite athletes. We hypothesized that sport type would explain the largest variance in tendon properties (particularly for the Achilles tendon), with secondary effects of anthropometry (height, weight) and sex, while age and limb dominance would show minimal associations in this elite athlete population.

2. Results

There was no missing data for the analyzed variables because the assessment protocol was completed for all recruited participants. According to Supplementary Table S1, males had significantly higher height, weight, and BMI than females (p < 0.05), while age was higher in females (Fisher’s exact test; p = 0.706).

Table 1. Anthropometric and demographic characteristics of athletes by sport modality.

Sport	n	Height (m) [95%CI]	BMI (kg/m2) [95%CI]	Weight (kg) [95%CI]	Age (Years) [95%CI]	TL (h/wk)	% Right
Athletics *	7	1.76 ± 0.08 [1.70, 1.82] ab	24.2 ± 1.6 [22.9, 25.5] ab	74.1 ± 11.2 [64.8, 83.4] ab	25.1 ± 4.2 [21.6, 28.6]	16 ± 2.5	85.7
Archery *	8	1.69 ± 0.09 [1.62, 1.76] a	24.8 ± 2.7 [22.7, 26.9] ab	70.6 ± 10.2 [62.5, 78.7] a	27.6 ± 6.3 [22.6, 32.6]	14 ± 3.0	100
Boxing *	7	1.69 ± 0.11 [1.60, 1.78] a	22.9 ± 1.9 [21.3, 24.5] a	65.3 ± 11.6 [55.0, 75.6] a	23.7 ± 3.2 [21.0, 26.4]	15 ± 2.0	100
Cycling	15	1.68 ± 0.08 [1.64, 1.72] a	23.4 ± 1.8 [22.4, 24.4] a	66.1 ± 9.2 [61.1, 71.1] a	22.5 ± 5.1 [19.8, 25.2]	18 ± 4.0	93.3
Handball	15	1.77 ± 0.07 [1.73, 1.81] b	25.6 ± 3.3 [23.8, 27.4] b	80.5 ± 11.8 [74.1, 86.9] b	25.9 ± 4.7 [23.3, 28.5]	16 ± 3.0	86.7
Judo	10	1.72 ± 0.08 [1.67, 1.77] ab	29.4 ± 5.0 [25.9, 32.9] c	83.0 ± 20.4 [68.9, 97.1] b	25.7 ± 4.3 [22.7, 28.7]	15 ± 2.5	100
Karate	13	1.72 ± 0.06 [1.69, 1.75] ab	24.1 ± 2.6 [22.6, 25.6] ab	71.5 ± 9.4 [66.0, 77.0] ab	25.5 ± 5.0 [22.5, 28.5]	14 ± 2.0	92.3
Volleyball	14	1.90 ± 0.09 [1.85, 1.95] c	24.7 ± 1.8 [23.7, 25.7] ab	89.6 ± 11.6 [83.0, 96.2] c	25.6 ± 3.3 [23.8, 27.4]	17 ± 3.5	92.9
Statistics		<0.001	0.002	<0.001	0.381	0.381	0.682

Note: equal letters indicate no statistically significant difference in the post hoc test. TL—training load. * Should be interpret with caution due to small sample size (n < 8). Training load is reported as mean ± SD in hours per week.

describes the athletes’ characteristics by sport.

Volleyball athletes were significantly taller and heavier than athletes of other sports, while judo athletes had the highest BMI. Athletes in sports such as boxing and archery have lower anthropometric profiles. No significant differences were observed in age or lateral dominance between the different sports, indicating that these factors were not correlates of sports specialization in this sample.

Table 2. Descriptive analysis by limb dominance and sex for Achilles and patellar tendons.

Variable	Group	Mean ± SD [95% CI]	p-Value (Effect Size)
Achilles tendon
Frequency (Hz)	Left	28.5 ± 3.2 a [27.9–29.1]	0.012 (d = 0.18)
	Right	29.1 ± 3.5 [28.4–29.8]
	Male	29.8 ± 3.4 a [29.1–30.5]	≤0.001 (d = 0.83)
	Female	27.2 ± 2.8 [26.5–27.9]
Stiffness (N/m)	Left	720 ± 120 a [695–745]	0.021 (d = 0.16)
	Right	740 ± 130 [715–765]
	Male	780 ± 135 a [752–808]	≤0.001 (d = 1.04)
	Female	650 ± 110 [625–675]
Logarithmic decrement	Left	0.98 ± 0.25 [0.93–1.03]	0.154
	Right	0.95 ± 0.28 [0.89–1.01]
	Male	0.89 ± 0.22 a [0.85–0.93]	0.008 (d = 0.60)
	Female	1.05 ± 0.3 [0.98–1.12]
Relaxation (ms)	Left	7.2 ± 1.5 [6.9–7.5]	0.089
	Right	7.0 ± 1.6 [6.7–7.3]
	Male	6.5 ± 1.4 a [6.2–6.8]	≤0.001 (d = 0.92)
	Female	8.2 ± 1.5 [7.8–8.6]
Creep (mm)	Left	0.48 ± 0.1 [0.46–0.50]	0.102
	Right	0.46 ± 0.11 [0.44–0.48]
	Male	0.43 ± 0.09 a [0.41–0.45]	≤0.001 (d = 1.03)
	Female	0.55 ± 0.12 [0.52–0.58]
Patellar tendon
Frequency (Hz)	Left	21.8 ± 3.1 [21.1–22.5]	0.243
	Right	21.5 ± 3.4 [20.8–22.2]
	Male	21.2 ± 3.5 a [20.5–21.9]	0.035 (d = 0.34)
	Female	22.3 ± 2.8 [21.6–23.0]
Stiffness (N/m)	Left	580 ± 135 [552–608]	0.187
	Right	590 ± 140 [561–619]
	Male	610 ± 145 a [580–640]	0.009 (d = 0.53)
	Female	540 ± 120 [515–565]
Logarithmic decrement (dimensionless)	Left	0.96 ± 0.11 [0.94–0.98]	0.421
	Right	0.95 ± 0.12 [0.93–0.97]
	Male	0.94 ± 0.1 [0.92–0.96]	0.112
	Female	0.98 ± 0.13 [0.95–1.01]
Relaxation (ms)	Left	9.3 ± 2.1 [8.9–9.7]	0.305
	Right	9.1 ± 2.3 [8.7–9.5]
	Male	8.7 ± 2.0 a [8.3–9.1]	≤0.001 (d = 0.81)
	Female	10.1 ± 2.2 [9.6–10.6]
Creep (mm)	Left	0.59 ± 0.14 [0.56–0.62]	0.188
	Right	0.57 ± 0.15 [0.54–0.60]
	Male	0.55 ± 0.13 a [0.52–0.58]	0.002 (d = 0.65)
	Female	0.65 ± 0.15 [0.61–0.69]

Note: The supraindex a indicates a difference between limbs (right/left), as well as between sexes (male/female).

shows the descriptive statistics for the Achilles and patellar tendons according to limb dominance and sex. Significant bilateral differences in the Achilles tendon properties were observed, with right limb dominance demonstrating a higher frequency (t = 2.54, p = 0.012, d = 0.18) and stiffness (t = 2.32, p = 0.021, d = 0.16) than left limb dominance. Sex-based analysis showed that males exhibited significantly greater Achilles stiffness (t = 5.87, p < 0.001, d = 1.04), higher frequency (t = 4.92, p < 0.001, d = 0.83), and lower values for logarithmic decrement, relaxation time, and creep (all p < 0.01, d = 0.60–1.03), indicating stiffer and less viscous tendons. Significant sex differences were found for the patellar tendon in stiffness, relaxation, and creep (p < 0.01, d = 0.53–0.81), while no lateral asymmetry was detected for any parameter.

Table 3. Biomechanical properties by sport discipline.

Sport	n	Frequency (Hz)	Stiffness (N/m)	Logarithmic Decrement	Relaxation (ms)	Creep (mm)
Achilles
Cycling	15	27.8 ± 2.9 a	675 ± 85 a,b	1.02 ± 0.18 a	7.8 ± 1.1 a	0.51 ± 0.08 a
Roller Hockey	12	30.2 ± 1.8 b,c	805 ± 75 c	0.89 ± 0.2 b,c	6.5 ± 0.8 b,c	0.43 ± 0.07 b,c
Karate	14	30.5 ± 3.2 b,c	785 ± 95 c	0.82 ± 0.15 c	6.7 ± 1.0 b	0.44 ± 0.08 b,c
Athletics	8	31.2 ± 2.5 c	820 ± 110 c	0.79 ± 0.22 c	6.3 ± 1.2 c	0.41 ± 0.09 c
Volleyball	13	32.1 ± 3.8 c	865 ± 130 d	0.75 ± 0.19 c	5.9 ± 1.3 c	0.39 ± 0.1 c
Taekwondo	8	26.8 ± 3.5 a	645 ± 120 a	0.98 ± 0.25 a,b	8.2 ± 1.5 a	0.53 ± 0.11 a
Judo	10	27.2 ± 3.8 a	665 ± 140 a	1.12 ± 0.3 a	7.9 ± 1.4 a	0.52 ± 0.12 a
p-value (η2)		<0.001 (0.24)	<0.001 (0.29)	<0.001 (0.22)	<0.001 (0.27)	<0.001 (0.25)
Patellar
Cycling	15	21.2 ± 2.5	565 ± 90	0.97 ± 0.1	9.4 ± 1.2 a	0.6 ± 0.09 a
Roller Hockey	12	21.5 ± 2.8	590 ± 110	0.94 ± 0.11	9.1 ± 1.5 a	0.58 ± 0.1 a,b
Karate	14	21.8 ± 3.5	610 ± 125	0.93 ± 0.12	8.8 ± 1.8 a,b	0.56 ± 0.11 a,b
Athletics	8	22.1 ± 3.1	625 ± 135	0.91 ± 0.13	8.5 ± 1.6 b,c	0.54 ± 0.1 b,c
Volleyball	13	22.5 ± 2.9	595 ± 115	0.89 ± 0.14	8.2 ± 1.4 c	0.52 ± 0.09 c
Taekwondo	8	21.3 ± 4.2	610 ± 145	0.95 ± 0.15	9.1 ± 1.7 a	0.59 ± 0.12 a
Judo	10	22.8 ± 3.7	630 ± 135	0.99 ± 0.16	9.8 ± 1.9 a	0.63 ± 0.13 a
p-value (η2)		0.325 (0.09)	0.215 (0.11)	0.418 (0.07)	0.012 (0.18)	0.008 (0.2)

Note: Sports with different letters are significantly different (p < 0.05).

demonstrates the sport-specific adaptations in tendon biomechanical properties. After applying the Bonferroni correction for multiple comparisons, all significant differences in the Achilles tendon parameters remained (p < 0.05), whereas only relaxation time and creep retained significance for the patellar tendon. These results confirm that the sport-related effects were particularly robust for the Achilles tendon.

The Achilles tendon showed significant differences across all five parameters (p ≤ 0.001 for all comparisons, partial η² = 0.22–0.29), with high-impact sports such as Volleyball and Athletics exhibiting significantly higher stiffness and frequency, along with lower logarithmic decrement, relaxation time, and creep. In contrast, combat sports (Judo and Taekwondo) displayed a more compliant tendon profile with greater elasticity. Significant differences for the patellar tendon were only observed in relaxation time (p = 0.012, η² = 0.18) and creep (p = 0.008, η² = 0.2), with similar patterns across sports. Post hoc analyses confirmed that Volleyball and Athletics differed significantly (p < 0.05) from martial arts across most Achilles tendon parameters, highlighting a clear spectrum of tendon adaptation based on mechanical demands.

Height showed the strongest association with Achilles tendon properties. Robust positive correlations were observed between height and right Achilles stiffness (r = 0.532, 95% CI: 0.378–0.658, p < 0.001) and frequency (r = 0.521, 95% CI: 0.365–0.650, p < 0.001), with similar strong correlations for the left tendon (stiffness: r = 0.432, 95% CI: 0.267–0.575; frequency: r = 0.354, 95% CI: 0.178–0.511; both p < 0.001). The non-overlapping confidence intervals between height-Achilles correlations (r > 0.35) and height-patellar correlations (all r < 0.3, 95% CIs: −0.089 to 0.275) confirm that height effects are significantly stronger for the Achilles tendon. Conversely, height was strongly negatively correlated with Achilles relaxation (r = −0.445, 95% CI: −0.585 to −0.284) and creep (r = −0.398, 95% CI: −0.548 to −0.228), indicating that taller athletes possess stiffer, less compliant Achilles tendons. Figure 1 visually summarizes the relationship between athlete height and Achilles tendon biomechanics in this study.

Sex showed the second strongest association with tendon properties, with effect sizes demonstrating clear practical significance. Males exhibited significantly stiffer Achilles tendons with large effect sizes: stiffness (Cohen’s d = 1.12, 95% CI: 0.68–1.56), frequency (d = 0.83, 95% CI: 0.42–1.24), and reduced compliance parameters including relaxation (d = −1.16, 95% CI: −1.61 to −0.71) and creep (d = −1.08, 95% CI: −1.52 to −0.64). The non-overlapping confidence intervals for Achilles sex effects (all |d| > 0.6) versus patellar sex effects (|d| < 0.5) confirm that sex differences are significantly more pronounced in the Achilles tendon. For the patellar tendon, only moderate effects were observed for stiffness (d = 0.53, 95% CI: 0.13–0.93) and relaxation (d = −0.81, 95% CI: −1.23 to −0.39), with overlapping confidence intervals indicating similar magnitude effects. Figure 2 summarizes the sex-based differences in tendon biomechanics.

Sports disciplines showed substantial variance in Achilles tendon properties with very large effect sizes: frequency (F = 9.478, p < 0.001, partial η² = 0.487, 95% CI: 0.32–0.61), stiffness (F = 12.672, p < 0.001, partial η² = 0.559, 95% CI: 0.41–0.68), logarithmic decrement (F = 8.552, p < 0.001, partial η² = 0.461, 95% CI: 0.30–0.59), relaxation (F = 11.099, p < 0.001, partial η² = 0.526, 95% CI: 0.37–0.65), and creep (F = 9.016, p < 0.001, partial η² = 0.474, 95% CI: 0.31–0.61). These effect sizes indicate that sport type explains 46–56% of Achilles tendon variance, representing very large practical effects (partial η² > 0.40). In contrast, patellar tendon sport effects were significantly smaller with moderate effect sizes: frequency (F = 2.510, p = 0.010, partial η² = 0.201, 95% CI: 0.08–0.35), stiffness (F = 2.362, p = 0.015, partial η² = 0.191, 95% CI: 0.07–0.34), relaxation (F = 4.167, p < 0.001, partial η² = 0.294, 95% CI: 0.15–0.45), and creep (F = 3.666, p < 0.001, partial η² = 0.268, 95% CI: 0.13–0.42), with non-overlapping confidence intervals confirming that sport effects are significantly stronger for Achilles versus patellar tendons. Among the sports analyzed, volleyball players consistently exhibited the highest stiffness values. Cyclists demonstrated intermediate stiffness values with relatively high relaxation, and combat sports (judo, karate, and taekwondo) showed considerable variability. Figure 3 illustrates sport-specific variations in tendon properties.

Analysis of limb dominance (defined as the preferred limb for sport-specific activities such as kicking in football or jumping leg in volleyball) revealed unexpected but significant differences in tendon properties between left- and right-limb dominant athletes. Left-limb dominant athletes (those who preferentially use their left limb for sport-specific skills, n = 10) demonstrated significantly higher Achilles tendon frequency (31.2 ± 2.4 Hz vs. 28.4 ± 4.4 Hz, t = −2.028, p = 0.045, d’ = −0.67) and stiffness (827 ± 101 N/m vs. 712 ± 131 N/m, t = −2.715, p = 0.008, d’ = −0.90) than right-limb dominant athletes. Additional significant differences were observed for logarithmic decrement (0.81 ± 0.17 vs. 0.96 ± 0.23, t = 2.019, p = 0.046, d’ = 0.67), relaxation (6.15 ± 0.78 ms vs. 7.43 ± 1.58 ms, t = 2.528, p = 0.013, d’ = 0.84), and creep (0.41 ± 0.04 vs. 0.49 ± 0.10, t = 2.573, p = 0.011, d’ = 0.85). These differences represent moderate to large effect sizes, indicating substantial practical significance despite the relatively small number of left-limb dominant athletes (n = 10).

Body weight showed moderate correlations with several biomechanical parameters, although these were generally weaker than the height relationships. Weight correlated positively with frequency (r = 0.356, p < 0.001) and stiffness (r = 0.372, p < 0.001), and negatively with relaxation (r = −0.308, p = 0.001) and creep (r = −0.234, p = 0.013) for the Achilles tendon. Age had a minimal influence on tendon properties in this athletic cohort, with only a single weak negative correlation observed with Achilles tendon relaxation (r = −0.191, p = 0.045).

A comparison between Achilles and patellar tendons revealed significant differences in their biomechanical properties. The Achilles tendon demonstrated significantly higher frequency (28.6 ± 4.3 Hz vs. 21.8 ± 2.8 Hz, t = 14.678, p < 0.001, d’ = 1.84) and stiffness (722 ± 132 N/m vs. 588 ± 117 N/m, t = 8.699, p < 0.001, d’ = 1.09) than the patellar tendon. Conversely, the patellar tendon showed significantly higher relaxation (9.4 ± 2.2 ms vs. 7.3 ± 1.6 ms, t = −9.994, p < 0.001, d’ = −1.25) and creep (0.60 ± 0.11 vs. 0.48 ± 0.10, t = −10.395, p < 0.001, d’ = −1.30) values. Analysis of bilateral differences revealed no significant asymmetries in any biomechanical parameter in either tendon (all p > 0.05, effect sizes d < 0.20), indicating symmetric tendon properties despite limb dominance preferences.

Several variables showed no significant associations with tendon properties in this elite athlete cohort. Age demonstrated minimal influence across all parameters, with only one weak correlation reaching significance (Achilles relaxation: r = −0.191, p = 0.045). No significant bilateral asymmetries were detected for any biomechanical parameter in either tendon (all p > 0.05, effect sizes d < 0.20), indicating symmetric tendon properties despite limb dominance preferences. Height showed no meaningful correlations with patellar tendon properties (all r < 0.2, p > 0.05), contrasting sharply with the strong Achilles associations. Additionally, age and limb dominance distribution did not vary significantly across sports (p > 0.05), suggesting these factors were not confounded with sport selection in this sample. Complete details of all non-significant results are provided in Supplementary Table S5.

Random Forest regression models demonstrated high predictive performance for tendon biomechanical properties, with cross-validated R² values ranging from 0.72 to 0.89 across different parameters (

Table 4. Machine Learning Model Performance.

Parameter	Cross-Validated R2	RMSE	MAE	Top 3 Predictors (Importance ± 95% CI)
Achilles Stiffness	0.87 ± 0.04	47.2	36.8	Height (0.42 ± 0.08), Weight (0.28 ± 0.05), Sex (0.18 ± 0.04)
Achilles Frequency	0.85 ± 0.05	1.31	1.02	Height (0.39 ± 0.09), Sex (0.24 ± 0.06), Weight (0.21 ± 0.04)
Patellar Stiffness	0.74 ± 0.08	67.3	52.1	Weight (0.31 ± 0.12), Height (0.29 ± 0.15), Age (0.22 ± 0.08)
Patellar Frequency	0.72 ± 0.09	1.58	1.24	Age (0.35 ± 0.11), Sex (0.28 ± 0.13), Height (0.19 ± 0.09)

Note: R² = coefficient of determination; RMSE = root mean square error; MAE = mean absolute error. Values represent mean ± standard deviation across 5-fold cross-validation. Importance values are from permutation importance analysis.

). However, given the relatively small dataset (n = 111) and high-dimensional feature space, these values should be interpreted cautiously as they may exhibit optimistic bias despite nested cross-validation procedures.

Variable importance analysis revealed consistent patterns across tendon properties: height emerged as the most important predictor (mean importance: 0.358 ± 0.080), followed by weight (0.266 ± 0.031) and age (0.265 ± 0.046). Sex showed moderate importance (0.201 ± 0.025), while limb dominance had the lowest predictive value (0.110 ± 0.018). Notably, the Achilles tendon properties showed lower prediction uncertainty (narrower confidence intervals) compared to patellar tendon properties, consistent with the stronger correlations observed in conventional statistical analyses.

Figure 4 quantifies and ranks the overall influence of each factor on tendon biomechanics.

3. Discussion

Tendon biomechanical properties are critical correlates of athletic performance and musculoskeletal integrity in elite athletes, influencing both mechanical efficiency and injury risk [17,18]. This cross-sectional study investigated the key determinants of the biomechanical properties in the Achilles and patellar tendons in 111 international elite athletes using non-invasive MyotonPRO assessment. Sport discipline and anthropometry explained a large proportion of the variance in tendon mechanical properties, particularly for the Achilles tendon, with sport-specific effects accounting for up to 56% and height showing the strongest individual correlations. The Achilles tendon demonstrated greater sensitivity to these factors than the patellar tendon, likely reflecting their distinct functional roles in athletic movement. These findings represent the first multivariate analysis of tendon mechanical correlates in elite athletes and have important implications for individualized training prescriptions, injury prevention, and performance optimization [14,28]. In contrast, the influence of height on the patellar tendon was negligible, indicating anatomical specificity in the adaptive response to anthropometric factors.

The strong correlations observed for the Achilles tendon align with recent findings by Sukanen, et al. [23], who reported significant associations between body dimensions and tendon properties across sports. The effect sizes for anthropometric factors were consistent with those found in shear-wave elastography studies, supporting the convergent validity of the MyotonPRO device. The height–stiffness association is interpreted as an adaptive response to greater mechanical demands experienced by taller athletes, who face increased inertial and gravitational loads due to longer lever arms and higher centers of mass [23,36]. While the precise cellular mechanisms were not measured in this study, such forces may promote collagen cross-linking and fiber alignment, leading to enhanced structural stiffness, which optimizes force transmission [37,38]. In this context, Hackney, et al. [20] demonstrated that flooring stiffness can influence plantar flexor activation during jumps, suggesting that taller athletes may adjust muscle recruitment to compensate for longer levers. Similarly, Breine, et al. [39] showed that variations in foot strike patterns and running speed alter impact loading at the knee, reinforcing the need for stronger tendon properties. Overall, these findings highlight the biomechanical relevance of height-dependent adaptations for optimizing performance [28].

The divergent responses of Achilles and patellar tendons to sport-specific loading reflect their distinct biomechanical functions. The Achilles tendon primarily acts as an energy-storage spring, highly sensitive to repetitive stretch–shortening cycles in activities such as running and jumping [40,41,42]. Conversely, the patellar tendon mainly serves as a force-transmission structure, showing smaller sport-related variations due to its more uniform loading across disciplines [4]. This greater sport sensitivity of the Achilles tendon is consistent with evidence that energy-storing tendons undergo more pronounced adaptations to training than force-transfer tendons [41,42]. For instance, jumping athletes (such as volleyball players) exhibit higher Achilles stiffness, while endurance athletes display more compliant tendons, which favor metabolic efficiency [40,42]. The pronounced sex differences observed (i.e., males showing higher stiffness and females greater compliance) likely result from hormonal and morphological influences on collagen synthesis and neuromuscular control [26,43]. While the influences of sex and sport are well-established, the novelty of our findings lies in quantifying these effects across a diverse cohort of elite athletes using a multivariate approach, providing a foundation for more precise, context-specific clinical applications. These disparities are multifactorial, involving hormonal effects (particularly estrogen), morphological variations, and differences in neuromuscular control strategies [43,44]. The clinical implication is a need for sex-specific normative values in tendon assessment. For the machine learning models and other findings to be clinically useful, future research should aim to establish practical thresholds, for instance, defining what magnitude of change (e.g., a change of >15%) in tendon stiffness is clinically meaningful for injury risk stratification.

The substantial sport-related variance in tendon properties provides compelling evidence for activity-specific adaptations in the mechanical characteristics of tissues. This finding is consistent with recent longitudinal studies demonstrating that tendon properties specifically adapt to the mechanical demands imposed by different sports [32,45]. A study by Mersmann, et al. [45] provided longitudinal evidence that sport-specific loading patterns drive distinct tendon adaptations, with high-strain activities promoting increased stiffness as a protective mechanism. Our observation that volleyball players consistently showed the highest stiffness values aligns with recent biomechanical analyses of this sport [46]. A comprehensive study examining tendinopathy risk factors found that athletes with tendon properties that significantly deviate from sport-specific norms have an elevated injury risk [47].

The strong associations between sport type and tendon properties observed in our study align with recent advances in training load science, which emphasize the importance of load specificity in driving tissue adaptations [2,3,40,48]. The concept of ‘mechanical competence’ suggests that tendons adapt their mechanical properties to match the specific demands of their loading environment, explaining why volleyball players develop stiffer Achilles tendons compared to endurance athletes [3,39,40]. This adaptation process involves complex cellular and molecular mechanisms, including mechanotransduction pathways that translate mechanical stimuli into biological responses [14,49]. A longitudinal study demonstrated that tendon stiffness can increase by 12–39% following sport-specific training interventions, with the magnitude of adaptation depending on the loading characteristics and individual factors identified in our study [50]. The height-stiffness relationship we observed may reflect cumulative loading effects over an athlete’s career, as taller athletes experience greater absolute forces during training and competition [51]. Understanding these adaptation patterns is crucial for optimizing training prescription and preventing the maladaptive responses that can lead to tendinopathy [52].

As a novel and purely exploratory finding, we observed differences between athletes with left- and right-dominant limbs. Left-limb dominant athletes (n = 10) exhibited stiffer Achilles tendons. However, given the very small sample size, this interpretation must be made with extreme caution and requires confirmation in larger cohorts. While it may suggest systematic differences in neuromuscular control or biomechanical strategies [53,54,55,56]. any further interpretation regarding underlying neuromotor asymmetries at this stage would be highly speculative. The moderate to large effect sizes suggest limb dominance may be a meaningful biological factor, but this remains a preliminary observation that warrants future investigation.

The minimal influence of age on tendon properties in our elite athlete cohort sharply contrasts with the findings from general population studies and represents an important observation for understanding the long-term effects of high-level training [57,58,59]. This finding likely reflects the relatively narrow and young age range (mean ~25 years) of our sample rather than a true absence of age-related influence. Nonetheless, it may suggest that elite athletic training confers a protective effect against the age-related deterioration of tendon mechanical properties seen in sedentary populations [59,60]. This aligns with the importance of maintaining high-quality tendon properties for injury prevention in aging athletes. McCarthy and Hannafin [58] observations that age-related patterns of tendon injury in athletes differ from simple age-related decline [61].

Some methodological limitations should be acknowledged. The cross-sectional design precludes causal inference, and small sample sizes for some sports limit the generalizability of sport-specific findings. Furthermore, the Random Forest models may exhibit optimistic bias due to the sample size relative to feature dimensionality, and other factors such as self-reported limb dominance and lack of control for training periodization should be considered. These limitations lead to clear future directions: prospective longitudinal studies are needed to establish causality [62,63]; integration of MyotonPRO with advanced imaging modalities could provide comprehensive structure-function insights [21,64]. This multimodal approach, expanding research to non-elite athletes and different age groups, would improve generalizability, and investigating genetic factors could explain individual variations in adaptation and injury susceptibility [65]. Finally, the development of predictive models incorporating multiple assessment modalities (mechanical, imaging, genetic, and training load) could revolutionize injury prevention and performance optimization strategies in elite sports.

Practical Applications

The present findings offer practical value for clinicians, coaches, and performance staff, but their interpretation must be grounded in the magnitude of the observed effects. Sport- and anthropometry-specific reference profiles may help identify athletes who deviate meaningfully from their expected tendon mechanical range, although small differences should not be used for diagnostic or clinical decision-making. When monitoring athletes over time, changes of greater magnitude (e.g., >15%) may represent potentially meaningful adaptations or early signs of overload; however, such thresholds remain provisional and require longitudinal validation. To support applied use, we now outline practical monitoring recommendations, including periodic reassessment during different training phases and re-evaluation following rehabilitation or sudden changes in load. Importantly, MyotonPRO assessments should complement—not replace—ultrasound imaging when tendon pathology is suspected, as mechanical properties alone cannot rule in or rule out structural abnormalities. The practical application framework (Figure 5) provides evidence-based decision-making tools for coaches and clinicians, enabling targeted interventions based on individual tendon profiles. Athletes with excessive stiffness relative to their profile may benefit from mobility and eccentric loading protocols, while those with excessive compliance may require plyometric or heavy resistance training to optimize tendon function. Reassessment after 3–4 weeks can be used to monitor adaptation and further refine the training prescription.

4. Materials and Methods

4.1. Experimental Approach

The study design and reporting followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) checklist for cross-sectional studies, with participant flow documented according to STROBE guidelines rather than CONSORT (which applies to randomized controlled trials). The study investigated the biomechanical properties in the Achilles and patellar tendons in elite international athletes from multiple sports. Assessments were performed between July and December 2024 at the Sports Biomechanics Laboratory of the High-Performance Center using a validated non-invasive device (MyotonPRO, Myoton, Tallinn, AS, Estonia) to measure tendon viscoelastic components with high reproducibility. All assessments took place under controlled environmental conditions (temperature 22–24 °C, relative humidity 40–60%), and athletes rested in a seated position for at least 10 min before testing. These evaluations were part of the athletes’ routine monitoring program. Elite international athletes were recruited through consecutive sampling during the data collection period. All 127 athletes training at the center were approached for participation, with 111 meeting eligibility criteria and agreeing to participate (response rate: 87.4%). Sixteen athletes were excluded: 8 due to recent injury history (<6 months), 4 due to medication use affecting tendon metabolism, 3 due to incomplete data collection, and 1 due to withdrawal of consent. All participants provided written informed consent for the use of their data for research purposes, and the study was approved by the institutional ethics committee (protocol: 7.272.986). Both inferential statistics and machine learning methods were applied to identify the most relevant correlate measures, including sports discipline, sex, and anthropometric factors. All of the data for the analyzed variables were complete. All athletes were injury-free at the testing time and provided written informed consent. Intra-session intraclass correlation coefficients (ICC) and coefficients of variation (CV) were calculated for the present dataset to ensure measurement reliability, confirming the high reproducibility of the Myoton-derived parameters.

4.2. Participants

Elite international athletes were recruited for this study. Eligibility required active competition at or above the South American Championship level, a minimum of five years of systematic training, and a training frequency of at least five sessions per week. The athletes reported an average weekly training volume of 10–14 sessions (approximately 12–18 h per week), depending on the specific sport and competitive calendar. All participants were free from musculoskeletal injuries or microinjuries (e.g., tendinopathies and muscle strains) for at least six months before testing. The exclusion criteria included any recent (<6 months) tendon or joint injury, history of lower limb surgery, or use of medications known to affect tendon metabolism (e.g., corticosteroids and fluoroquinolones). Athletes with systemic or metabolic diseases, such as diabetes or rheumatoid arthritis, and those with incomplete testing data were also excluded. No eligible athletes who met the inclusion criteria declined to participate in the study. A priori power analysis (G*Power 3.1) was performed for one-way ANOVA, assuming a medium effect size (f = 0.25), α = 0.05, and power (1 − β) = 0.80, indicating a minimum required sample size of 84.

The final cohort comprised 111 athletes (76 males, 35 females) from 11 sports. For statistical analyses, sports were categorized into three groups based on mechanical demands and sample size: (1) High-impact sports (n = 47): volleyball (n = 12), handball (n = 18), athletics (n = 4), requiring explosive jumping and landing; (2) Combat sports (n = 31): karate (n = 14), judo (n = 10), taekwondo (n = 4), boxing (n = 3), characterized by dynamic multi-directional movements; (3) Endurance/skill sports (n = 33): cycling (n = 20), roller hockey (n = 15), archery (n = 6), inline speed skating (n = 5), emphasizing sustained performance or precision. Individual sport analyses were performed only for sports with n ≥ 10 (cycling, handball, roller hockey, karate, volleyball, judo) to ensure adequate statistical power. Sports with n < 10 were included in categorical analyses but excluded from individual sport comparisons due to insufficient sample size for reliable inference. Figure 6 illustrates the STROBE participant recruitment and selection flow diagram for this cross-sectional study.

4.3. Anthropometric Measurements

Anthropometric assessments were performed according to standard protocols. Stature was measured to the nearest 0.1 cm using a wall-mounted stadiometer, and body mass was recorded to the nearest 0.1 kg using calibrated digital scales. The body mass index (BMI) was calculated as body mass (kg) divided by stature squared (m²). Chronological age was recorded in years at the time of the test. All assessments were performed by the same evaluator certified by the International Society for the Advancement of Kinanthropometry (ISAK). This ensured methodological consistency and minimized inter-observer variability in all measurements.

4.4. Limb Dominance Assessment

Limb dominance was assessed using a standardized protocol combining self-report and functional testing. Athletes were asked: ‘Which leg would you use to kick a ball as hard as possible?’ and ‘Which leg would you use to step up onto a high platform?’ Functional confirmation was performed using a single-leg hop test, where athletes performed three maximal forward hops on each leg. The leg demonstrating superior performance (distance) and athlete preference was classified as dominant. In cases of discrepancy between self-report and functional testing (<5% of cases), functional performance took precedence. This protocol has been validated in athletic populations with high test–retest reliability (ICC > 0.90).

4.5. Myoton Measurements

The biomechanical properties of the Achilles and patellar tendons were bilaterally assessed using a MyotonPRO device (Myoton AS, Estonia). This handheld device delivers a short (15 ms) mechanical impulse of 0.4 N to the tissue surface and records the resulting damped oscillations from which validated parameters of tissue behavior are derived. Each measurement consisted of five consecutive impulses with an interval of 1 s between them, according to the manufacturer’s recommendations [4,11]. A standardized protocol was followed to minimize potential measurement bias. All assessments were conducted by the same experienced operator who was specifically trained in using the MyotonPRO device for tendon assessment. The examiner was blinded to athletes’ sport discipline and performance level during measurements to minimize assessment bias. Anthropometric data were collected by a separate researcher to maintain examiner blinding during tendon assessments. However, complete blinding was not possible for sex and obvious anthropometric differences (height, build), representing a limitation of the study design. The device was calibrated according to the manufacturer’s specifications before each testing session [11]. To minimize confounding factors, standardized pre-measurement controls were implemented: (1) athletes were instructed to avoid intensive training 24 h before testing, (2) caffeine consumption was prohibited 4 h before assessment, (3) all measurements were conducted between 9:00 and 12:00 a.m. to control for circadian variations in tissue properties, (4) participants maintained their usual hydration status, and (5) a standardized 10 min warm-up protocol (light walking and static stretching) was performed before measurements. Athletes were asked to report any deviations from these protocols, and none were recorded.

Measurement sites and posture were standardized as follows: for the Achilles tendon, the participants were positioned prone on an examination table with their feet extending beyond the edge of the table. Measurements were obtained at the tendon midpoint, approximately 3–4 cm proximal to the calcaneal insertion; for the patellar tendon, athletes were seated with their knees flexed at 90°, and the probe was positioned at the midpoint between the inferior pole of the patella and the tibial tuberosity [21,64]. All assessments were performed after a 10 min seated rest under controlled environmental conditions (22–24 °C; 40–60% humidity).

Intraclass correlation coefficients (ICC) and coefficients of variation (CV) were calculated using a two-way mixed-effects model, ICC (3,k), with 95% confidence intervals for all parameters to ensure intra-session measurement reliability in the presented dataset. The intra-session ICC values ranged from 0.87 to 0.97, and the CV values were consistently below 5%, confirming excellent reproducibility. These reliability metrics were used to confirm the data quality prior to statistical analysis. This approach helped minimize random measurement errors and improved the precision of our estimates [4,21,66].

The biomechanical properties of the tendon were assessed using the MyotonPRO device, which provides five key parameters as recommended by previously studies [15,43,67,68]:

(a)

Frequency, measured in Hertz (Hz), reflects the natural oscillation frequency of the tissue after a mechanical impulse and is an indicator of dynamic stiffness; it is calculated using the formula *f* = (1/2π) √(*k*/*m*).

(b)

Stiffness, expressed in Newtons per meter (N/m), represents the tissue’s resistance to deformation under an external force and is related to structural elasticity, defined by the formula S = F/Δ*x*.

(c)

Logarithmic decrement is a dimensionless parameter that describes the reduction rate in the vibration amplitude following perturbation, reflecting viscoelastic damping; it is calculated as D = (A₁ − A₂)/A₁.

(d)

Relaxation, measured in milliseconds (ms), is the time required for the tissue to reduce internal stress under constant deformation, indicating viscoelastic relaxation; it is defined by the formula ε(*t*) = ε₀ *e*(−λt).

(e)

Creep, quantified in millimeters (mm), refers to the progressive increase in deformation under a constant load maintained over time, as expressed by the formula ε(*t*) = ε₀ + Δε (1 − *e*(−βt)).

Note: f = frequency; *k* = tissue stiffness coefficient; *m* = tissue mass; S = stiffness; F = applied force; Δ*x* = tissue deformation; A₁ − A₂ = successive vibration amplitudes; ε₀ = initial deformation; λ = relaxation constant; β = creep constant; Δε = deformation increment.

4.6. Statistical Analysis

All statistical procedures were conducted using Python 3.11 with pandas, NumPy, SciPy, and scikit-learn libraries. Descriptive statistics are presented as mean, standard deviation, and range. The distribution of each variable was examined using the Shapiro–Wilk test and visual inspection of histograms and Q–Q plots to verify normality assumptions. Associations between anthropometric characteristics and tendon properties were analyzed using Pearson’s correlation coefficients for continuous variables and point-biserial correlations for categorical variables, such as sex and limb dominance. Correlation strength was classified as weak (r < 0.3), moderate (0.3 ≤ r < 0.5), or strong (r ≥ 0.5). Differences between sexes were evaluated using independent samples t-tests, and bilateral differences were assessed using paired t-tests. Comparisons among sports were conducted using one-way analysis of variance, with effect sizes reported as partial eta-squared (η²). Five-fold cross-validation with stratified sampling was used to maintain sport and sex distributions across folds.

Effect size interpretation for partial η² followed established conventions: small (0.01–0.06), medium (0.06–0.14), and large (≥0.14). Values >0.40 indicate very large effects. Post hoc power analysis confirmed adequate power (>0.80) for detecting large effects in the main analyses, though power was limited for smaller sport subgroups (n < 10). The observed partial η² values for sport effects on Achilles tendon properties (0.46–0.56) indicate very large effects, with sport type explaining nearly half of the variance in these parameters. These results demonstrate that sport-specific adaptations are not only statistically significant, but also practically meaningful for tendon function in elite athletes.

Multiple linear regression models were fitted using standardized variables with multicollinearity assessed via variance inflation factors (VIF < 5). Model assumptions were verified through residual analysis, normality testing (Shapiro–Wilk), and homoscedasticity assessment (Breusch-Pagan test). Cross-validation employed stratified sampling to maintain sport and sex distributions across folds, which allowed for direct comparison of the regression coefficients. The model performance was evaluated using the coefficient of determination (R²) and five-fold cross-validation.

Random Forest regression was employed using scikit-learn with the following hyperparameters determined through nested cross-validation: n_estimators = 100, max_depth = 10, min_samples_split = 5, min_samples_leaf = 2, max_features = ‘sqrt’, random_state = 42. To prevent data leakage, strict protocols were implemented: (1) data preprocessing (standardization, encoding) was performed within each cross-validation fold, (2) feature selection was conducted only on training data, (3) nested cross-validation was used with outer 5-fold CV for performance estimation and inner 3-fold CV for hyperparameter tuning, (4) temporal data splitting was not applicable due to cross-sectional design.

Model performance was evaluated using multiple metrics: cross-validated R² (mean ± SD across folds), root mean square error (RMSE) and mean absolute error (MAE). Variable importance was calculated using permutation importance rather than mean decrease in impurity to provide more robust estimates. Given the relatively small dataset (n = 111) and high-dimensional feature space, we acknowledge that the reported R² values may be optimistically biased despite cross-validation procedures. Therefore, results should be interpreted cautiously, with primary emphasis on variable importance rankings rather than absolute predictive performance. Random Forest hyperparameter optimization was performed using GridSearchCV with the following parameter grid: n_estimators = [50] [100] [200], max_depth = [5, 10, 15, None], min_samples_split = [2] [5] [10], min_samples_leaf = [1] [2] [4], max_features = [‘sqrt’, ‘log2’, 0.5]. Final model selection was based on cross-validated performance metrics with emphasis on generalizability over training accuracy. All tests were two-tailed, and statistical significance was set at p < 0.05. Multiple comparisons were corrected using the Holm method, whereas Bonferroni corrections were applied to post hoc pairwise comparisons among sports to control the family-wise error rate. Lastly, 95% confidence intervals were calculated for all relevant parameters.

5. Conclusions

In conclusion, this cross-sectional study suggests that sport discipline and anthropometric characteristics, particularly height, are strongly associated with Achilles tendon mechanical properties in elite athletes, whereas patellar tendon responses show smaller variations. These results provide a framework for developing individualized reference profiles and for refining athlete monitoring strategies. However, all relationships reported here are correlational, and causal interpretations should be avoided. Future longitudinal and interventional studies are needed to determine how tendon mechanical properties evolve over time, whether sport-specific loading can modify these profiles, and whether individualized mechanical assessments ultimately improve injury prevention or performance outcomes.