**3. Results**

#### *3.1. Fatigue Protocol*

After the fatigue protocol, the mean values of HR, BL, and RPE were greater than 185 beats/min, 14 mmol/L, and a score of 18, respectively. Results of repeated measurement indicated that HR, BL, and RPE were all affected by the fatigue protocol (partial η2: 0.99, 0.876, and 0.994 respectively), and the *p* values were all less than 0.001 (Table 1).

Figure 1 illustrates the HR–time curve (mean) of four participants throughout the fatigue protocol. To divide the fatigue protocol into stages, time was normalized, the slope of the HR–time curve was calculated, and four stages were distinguished, namely, P1 (0–10% duration of fatiguing process (D)), P2 (10–30%D), P3 (30–60%D) and P4 (60–100%D).



**Figure 1.** Heart rate (HR) of four participants during the fatiguing process (mean). According to the slope of the HR–time curve, the overall duration of the fatiguing process was divided into four stages, that is, P1, P2, P3, and P4. Duration of fatiguing process (%) vs. heart rate (beats/min).

#### *3.2. Kinematics of Lower Extremity Joints*

In the sagittal plane, Table 2 shows that the hip, knee, and ankle initial contact angle (CAH, CAK, CAA, respectively), peak angle (PAH, PAK, PAA, respectively), time to peak angle, and range of motion (RoM) all decreased during the process of the fatigue protocol (from P1 to P4).

**Table 2.** Kinematics variables in the four stages of fatiguing process (P1–P4) from pre- to post-fatigue protocol in the sagittal plane (mean ± SD).


Notes: CAH, CAK, and CAA: initial contact angle of hip, knee, and ankle joint, respectively; PAH, PAK, and PAA: peak joint angle of hip, knee, and ankle joint, respectively. P1–P4: the 1st, 2nd, 3rd, and 4th stages of the fatiguing process, respectively. *a*, *a\**: significant differences between P1 and P2 at 0.05 and 0.01 levels, respectively; *b*: significant differences between P1 and P3 at 0.05 level; *c*: significant differences between P1 and P4 at 0.05 level; *d\**: significant differences between P2 and P3 at 0.01 level; *e\**: significant differences between P2 and P4 at 0.01 level; *f*: significant differences between P3 and P4 at 0.05 level.

At initial contact, a significant decrease was found for hip flexion (P1 vs. P4, *p* = 0.044; P2 vs. P3, *p* = 0.009; P2 vs. P4, *p* = 0.003) and knee flexion (P1 vs. P2, *p* = 0.028; P1 vs. P3, *p* = 0.037; P1 vs. P4, *p* = 0.004). Hip, knee, and ankle joints were all flexed (plantar-flexed) to the peak joint angle in less time. There were statistically significant differences in the time to peak value for the knee joint (P1 vs. P4, *p* = 0.009; P3 vs. P4, *p* = 0.012) and the ankle joint (P1 vs. P2, *p* = 0.036; P1 vs. P3, *p* = 0.021; P1 vs. P4, *p* = 0.001). The significance values (*p*) of the knee joint RoM were 0.013 (P1 vs. P2) and 0.048 (P1 vs. P4).

#### *3.3. Five Sub-Phases of the Stance Phase*

Table 3 shows that significant differences existed in the duration of the sub-stance phases II, III, IV, and V during the fatiguing process: durations were shorter for II (P1 vs. P3: *p* = 0.012), III (P1 vs. P4: *p* = 0.041; P3 vs. P4: *p* = 0.005), and V (P1 vs. P3: *p* = 0.021; P1 vs. P4: *p* = 0.005), and longer for IV (P1 vs. P2: *p* = 0.022; P1 vs. P3: *p* = 0.024; P1 vs. P4: *p* = 0.002; P2 vs. P4: *p* = 0.002; P3 vs. P4: *p* = 0.036).

#### *3.4. Kinetics of Lower Extremity Joints*

For the mean moments and power at the hip, knee, and ankle joints in the sagittal plane. Significant differences were found for moments at the hip (the 1st peak, P1 vs. P4: *p* = 0.03; the 2nd peak, P1 vs. P4: *p* = 0.01) and the knee (the 1st peak, P1 vs. P4: *p* < 0.001; the 2nd peak, P1 vs. P4: *p* = 0.01); power generation at the hip (P1 vs. P3: *p* = 0.025; P1 vs. P4: *p* = 0.014) and knee (P1 vs. P2, P1 vs. P3, P1 vs. P4: *p* < 0.001); and power consumption at the knee (P1 vs. P2: *p* = 0.029; P1 vs. P3: *p* = 0.009 and P1 vs. P4: *p* < 0.001) (details in Figure 4a).

#### *3.5. SPM (1D) of Kinematics of Lower Extremity Joints*

For the four stages of the fatiguing process (P1–P4), Figure 2a illustrates the joint angle–time curves in the sagittal plane (mean ± SD). SPM (1D) analysis results with significant differences are illustrated in Figure 2b. At the hip, significant differences exist in the initial contact (I), impact loading (II), weight acceptance (III), and pre-drive-off (IV) phases (P2 vs. P3 (I and II: *p* = 0.01; III and IV: *p* < 0.001); P2 vs. P4 (I–IV: *p* < 0.001)). At the knee, significant differences were found mainly in the weight acceptance and pre-drive-off phases (P1 vs. P2 and P1 vs. P4: *p* < 0.001).


**Table 3.** Duration of five sub-phases (%) (mean ± SD).

Notes: PF1: initial impact peak (the 1st peak vertical ground reaction force, vGRF); PF2: secondary impact peak (the 2nd peak vGRF); PF3: third impact peak (the 3rd peak vGRF); PAK: peak knee joint flexion angle. FO: foot <sup>o</sup>ff. The stance phase was divided into five sub-phases (I–V) according to impact peak and PAK. P1–P4: the 1st, 2nd, 3rd, and 4th stages of the fatiguing process, respectively. *a*, *a\**: significant differences between P1 and P2 at 0.05 and 0.01 levels, respectively; *b*: significant differences between P1 and P3 at 0.05 level; *c*, *c\**: significant differences between P1 and P4 at 0.05 and 0.01 levels, respectively; *e\**: significant differences between P2 and P4 at 0.01 level; *f*, *f\**: significant differences between P3 and P4 at 0.05 and 0.01 levels, respectively.

**Figure 2.** (**a**) Summary of the joint angles (mean ± SD) of lower limbs in the sagittal plane. (**b**) Results with significant differences of one-dimensional statistical parametric mapping (SPM (1D)) for hip and knee joint angles. Positive angles represent hip and knee flexion and ankle dorsi-flexion. P1–P4: the 1st, 2nd, 3rd, and 4th stages of the fatiguing process, respectively.

#### *3.6. SPM (1D) of Kinetics of Lower Extremity Joints*

Figure 3 illustrates the vGRF waveform data of P1, P2, P3, and P4, and the results of SPM (1D) analysis with significant differences, which are shown mainly in the pre-drive-off phases between P1 and P4 (*p* < 0.001).

Figure 4a illustrates the waveform data of the hip, knee, and ankle joint moments and power in the sagittal plane (mean ± SD) in P1, P2, P3, and P4. SPM (1D) analysis results of moments and power with significant differences are illustrated in Figure 4b,c, respectively. For hip and knee moments, significant differences exist in the partial period of the pre-drive-off phase (IV) between P1 and P3 (*p* < 0.001), and in the IV phase between P1 and P4 (*p* < 0.001). Moreover, Figure 4c illustrates that significant differences exist in the partial IV and drive-off (V) phases for hip power (P1 vs. P4: *p* < 0.05) and knee power (P1 vs. P3: *p* < 0.001; P1 vs. P4: *p* < 0.001).

**Figure 3.** Vertical ground reaction force (vGRF) and results with significant differences of one-dimensional statistical parametric mapping (SPM (1D)). P1–P4: the 1st, 2nd, 3rd, and 4th stages of the fatiguing process, respectively.

**Figure 4.** *Cont.*

**Figure 4.** (**a**) Summary of joint moments and power (mean ± SD) of lower extremities in the sagittal plane during the lunge. (**b**) Results with significant differences of one-dimensional statistical parametric mapping (SPM (1D)) for hip and knee joint moments. (**c**) Results with significant differences of SPM (1D) for hip and knee joint power. Positive moments represent hip and knee extensor and ankle plantar flexor moments, and positive joint power indicates periods of power generation. P1–P4: the 1st, 2nd, 3rd, and 4th stages of the fatiguing process, respectively. *a*, *a\**: significant differences between P1 and P2 at 0.05 and 0.01 levels, respectively; *b*: significant differences between P1 and P3 at 0.05 level; *c*, *c\**: significant differences between P1 and P4 at 0.05 and 0.01 levels, respectively.

## **4. Discussion**

An increasing number of people now play badminton. Both athletes and recreational players attempt to optimize their performance, thus increasing the risk of injury. Prevention of sports-related injuries is an important challenge. Fatigue is a major factor causing injury. Consequently, this study investigated the fatigue effects on a specific movement in badminton, i.e., the footwork associated with the lunge, which is one of the most used and integral movements [10,11].

Considering the task dependency of fatigue, a repeated forehand forward lunge, until reaching exhaustion, was proposed as the fatigue protocol, which was subdivided into four stages (P1, P2, P3, and P4) according to the slope of the heart rate (HR)–time curve. After the protocol, the mean values of HR, blood lactate (BL), and rating of perceived exertion (RPE) increased significantly at the significance level of 0.001; in particular, the mean HR was greater than 185 beats/min, the mean BL value was greater than 14 mmol/L, and the mean RPE score was greater than 18, indicating that all participants were fatigued. At P1, the first stage of the fatigue protocol, all participants were in a pre-fatigue state and, at P4, the final stage of the fatigue protocol, they were fatigued. In addition, according to previous studies [10,20,21], the first, second, and third impact peak (PF1, PF2, and PF3) and knee flexion peak angle (PAK) were used to subdivide the lunge stance phase into initial contact (I: 0-PF1)), impact loading (II: PF1-PF2), weight acceptance (III: PF2-PAK), pre drive-o ff (IV: PAK-PF3), and drive-o ff (V: PF3-end) periods. Then, statistical analyses were undertaken for both discrete and waveform kinematic and kinetic data in the sagittal plane, in which the largest movements occurred, comparing not only the pre- and post-fatigue states, but also the four stages of the fatiguing process.

The results supported the hypothesis that the biomechanical characteristics of the lunge change significantly during the fatiguing process. At the initial contact time, participants exhibited a more "erect" posture for the lunge, which is usually observed by coaches, and was shown as less hip and knee flexion, and ankle dorsi-flexion. Less dorsi-flexion at the foot strike has been found for recreational players with a relative lack of muscle power compared to national-level badminton athletes [19]. In the lateral jump performed in badminton, Herbaut et al. [27] found a decreased plantar-flexion angle at the foot strike post-fatigue. These changes may be induced by muscle fatigue caused by repeated stretch-shortening. Furthermore, the range of motion (RoM) of hip, knee, and ankle joints was decreased due to the fatiguing process.

Smaller peak joint angles were found at hip, knee, and ankle joints in the sagittal plane, particularly at the knee joint (with significant di fferences between P1 and the other three stages of the fatiguing process). After fatigue, a decrease in the knee peak angle was also found by Valldecabres et al. [16], however, no significant di fference was found. The discrepancy may be due to the participant's sports level [24]. The decrease in the peak angle could be explained by the decrease in joint moments (details in Figure 4). There were significant di fferences in peak joint moments at the hip and knee extensor (P1 vs. P4: hip (*p* = 0.03), knee (*p* < 0.001)). A similar relationship was illustrated by Fu et al. [19] between professional and amateur badminton players; that is, professional players with greater muscle strength and better performance showed higher knee and ankle joint moments. Additionally, a shorter time was taken to flex to the peak angle, with significant decreases in time shown at the knee and ankle. This may be caused by impaired control due to fatigue. This can also be explained in this study by the decrease in joint moments and power.

In addition to the angles, the durations of the four sub-stance phases (II–V) changed during the process of the fatigue protocol. The shorter durations of II and III indicate that less time was taken to reach PF2 and PAK, respectively; that is, due to the fatiguing process, the participant placed his foot flat and flexed his knee more quickly. This is consistent with the increased ankle plantar flexor and knee flexor moments in the present study (Figure 4a), and with the opinion that fatigue reduces the capacity of muscles to generate force. The most significant increase was found in the IV phase (pre-drive-o ff), increasing from 27.25 ± 10.69% stance to 43.38 ± 15.24% stance. Significantly decreased hip and knee peak power in this phase provided su fficient support for this change. Less power was generated for players for the drive-o ff. Kuntz et al. [10] indicated that a hop style lunge generates higher peak vertical force during loading. In this study, three participants used the hop style at the final stage of the fatigue protocol. The change of lunge style may be a strategy to generate more power for driving-o ff and returning to the starting position [14].

Furthermore, considering the time continuity of biomechanical variables of the lunge motion, one-dimensional statistical parametric mapping SPM (1D) was used to analyze the biomechanical waveform data in the sagittal plane. Most of the joint angle, moment, and power waveforms decreased consistently during the process of the fatigue protocol (among P1, P2, P3, and P4). Significant di fferences were found in the hip and knee joint angles, mainly in the pre-drive-o ff phase (IV) (most *p* values were less than 0.001). The results support the hypothesis that the fatigue e ffects were di fferent in the five

periods of the lunge. Moreover, these results support the view that lunge characteristics change due to fatigue.

Taking into account the joint moments, both the hip and knee joints showed a significant response to the fatigue protocol. Significant di fferences were found between P1 and P3 in part of the period of IV, and between P1 and P4 in IV. The hip moments had a larger e ffective scope. However, there were only significant di fferences in the hip power between P1 and P4. For knee power, the significant di fferences were seen between P1 and P3, and also between P1 and P4, with a larger e ffective scope of the period. This indicates that the significant decrease in joint power occurred earlier and was mostly evident at the knee; that is, the fatigue responses manifested mostly at the knee joint. This result is consistent with a previous epidemiology study that reviewed musculoskeletal injuries among Malaysian badminton players, which reported that the majority of injuries sustained by players were due to overuse, primarily of the knee [7]. Another study [28] suggested that the rapidly changing eccentric/concentric work of the quadriceps in the varying degree of knee flexion was probably associated with patellar tendon injury. This may be why patellar tendinopathy is the most common injury of lower limbs among badminton players, and this result also supports our third hypothesis. Coaches and players should pay more attention to the training of the knee, and particularly the knee extensors.

Additionally, a comparison indicates that the discrete and waveform results are consistent. There are significant di fferences in CAH (P2 vs. P3), CAK (P2 vs. P4), PAK (P1 vs. P2; P1 vs. P4), knee peak moment (P1 vs. P4), hip peak power (P1 vs. P4), and knee peak power (P1 vs. P3; P1 vs. P4). In addition, significant di fferences were shown in these periods of stance phase in the discrete data. The results of SPM (1D) analysis clearly provided more information, indicating the period rather than a point of time in the fatigue response. This is also helpful in identifying the key body segments a ffected by fatigue. Thus, it is important for coaches and players to design a corresponding training program to improve the technique and muscle strength. Moreover, the results of SPM (1D) can provide support for monitoring of training.

Considering the key findings of this study, a few limitations should be noted when interpreting the results. First, the participants were male badminton players with at least 8 years of special badminton training. Results may di ffer for players of di fferent levels, ages, and gender, thus studies of players with a range of abilities and ages, including female players, should be considered. Second, all tests were conducted on a simulated badminton court. Third, the sample size was limited. Fourth, the changing slope of the HR–time curve was used to subdivide the fatiguing process. However, it is not su fficient to explain the status of fatigue. Electromyography (EMG) data might be more suitable and could be used in future work. Finally, although joint moments and power allow further assessment of the functional contribution of the joints, EMG and musculoskeletal system simulations would help understand movement changes during the fatiguing process.
