Athlete monitoring strategies are used to understand imposed training loads, and to evaluate an athlete’s response to training stimuli. Monitoring strategies can be useful in optimizing an athlete’s performance by determining their position on the recovery-adaptation continuum following training exposures, managing training loads to mitigate injury risk, as well as establishing quantitative parameters to guide return-to-play and return-to-performance protocols following an injury (Halson, 2014; Bourdon et al., 2017; Dunlop et al., 2019; Taberner et., 2019). Monitoring external training load (eTL) refers to the assessment of mechanical or locomotive work completed by the athlete and provides sport performance coaches with an objective measure of work performed during training, as well as games (Halson, 2014; Heishman et al., 2018a; 2018b; Fox et al., 2018; Svilar et al., 2018a; 2018b). Wearable microsensors, known as inertial measurement units (IMUs) offer a practical and convenient option to quantify eTL in indoor team sports, such as basketball (Holme, 2015; Fox et al., 2017).

IMUs have been used to characterize eTL among basketball athletes during both practice and competition, with PlayerLoad™ (PL) frequently reported as the key workload variable (Scanlan et al., 2014; Schelling and Torres, 2016; Aoki et al., 2017; Heishman et al., 2017; 2018a; 2018b; Peterson and Quiggle, 2017; Fox et al., 2018; Svilar et al., 2018a; 2018b). PL is a vector of magnitude, expressed as the square root of the sum of the squared instantaneous rate of change in acceleration in each of the 3 orthogonal planes, divided by the scaling factor 100 and expressed in arbitrary units (au) (Barrett et al., 2014; Heishman et al., 2018a; 2018b). The computation of PL includes the summation of load vectors in all 3 orthogonal planes (mediolateral, anteroposterior, and vertical), however laboratory evidence suggests the vertical component of PL contributes 50-60% of load accumulation, while the mediolateral and anteroposterior components only contribute 20-25% of load accumulation during PL analysis (Barrett et al., 2014). Field-based analyses have identified strong correlations between PL and total distance traveled, suggesting the sensitivity of PL to running based activity, likely resulting from increased vertical accelerations from ground reaction forces during the gate cycle (Cormack et al., 2013; Barrett et al., 2014; Polgaze et al., 2015). McLean et al. (2018) speculated that an abundance of vertical acceleration data may mask smaller increases in mediolateral and anteroposterior vector activity, which may be pertinent to the sensitivity of eTL quantification. Moreover, it may be speculated that the large vertical component of basketball play (Schelling and Torres, 2016; Stojanović et al., 2018) could exacerbate the suppression of small increases in mediolateral and anteroposterior movements, such as increases associated with change-of-direction (CoD) activity. These findings have spawned contemporary interests among practitioners to determine alternative strategies for quantifying cumulative movement, such as 2-Demensional PL (PL^2D), that only includes the mediolateral and anteroposterior movements, as well as the evaluation of the each individual PL vector, however these analysis have yet to be performed in basketball.

In addition to quantitating work performed during training, some athlete monitoring strategies are used to evaluate the response of the athlete to the training imposed. The countermovement jump (CMJ) is commonly used to evaluate neuromuscular readiness and performance in sport (Aoki et al., 2017; Rowell et al., 2017; Heishman et al., 2018a; 2018b; Ferioli et al., 2018) and may provide insight regarding the capacity of an athlete to recover from training. Interestingly, previous research has reported increases (Aoki et al., 2017) and decreases in CMJ height over the preseason among professional basketball players, (Ferioli et al., 2018) decreases in collegiate players, (Heishman et al., 2017) while semi-professional athletes have revealed no change (Ferioli et al., 2018). These results may reflect the level of play or varying levels of eTL, which often go unquantified. Previous basketball literature has evaluated changes in CMJ height, while evidence from alternative sports suggests that different force-time characteristics may accentuate fatigue by identifying compensations in movement strategy to achieve the desired gross output (Cormack et al., 2008; Gathercole et al., 2015; Rowell et al., 2017). Of note, Flight Time to Contraction Time Ratio (FT:CT) evaluates the athletes’ jumping strategy and has recently been established as a reliable variable in collegiate basketball players (Heishman et al., 2018a; 2018b; 2019). Additionally, Reactive Strength Index Modified (RSI_Mod), derived from dividing contraction time (CT) by jump height, provides an index of explosiveness (Kipp et al., 2016), and may also be a useful parameter to quantify changes in performance (McMahon et al., 2018; Heishman et al., 2019). Therefore, coupling eTL with resultant changes in CMJ performance may allude to the dose-response relationship of training. Although an acute inverse relationship between eTL and subsequent CMJ performance has been established (Heishman et al., 2018a; 2018b; Cruz et al., 2018), no data exists paralleling eTL with CMJ performance in basketball athletes.

Limited data exist identifying the influence of player position on eTL in collegiate basketball players. Similarly, no data exist examining the impact of a player’s scholarship status, which alludes to their role on the team, on eTL. Furthermore, a paucity of literature is available relating eTL parameters with subsequent changes in CMJ performance parameters. Therefore, the purpose of the present study was to 1) characterize the average eTL per ses-Sion; 2) examine differences in the average eTL per session each week; and 3) explore changes in CMJ performance across the 5 weeks of preseason training phase in NCAA Division I basketball athletes. Subsequent analyses examined the influence of position and academic status on eTL. It was hypothesized that the average eTL per session would be similar across training weeks and that there would be a decrease in the neuromuscular performance indices of jump height, FT:CT, and RSI_mod across the preseason.

Fourteen male (age = 19.7 F 1.0 years, height = 1.98 0.07 m, body mass = 94.7 6.2 kg) NCAA Division I collegiate basketball players were included in this study. Participants were categorized into position groups consisting of forwards/centers (n = 7) or guards (n = 7) determined by the basketball coaching staff. Players were classified by academic status as either a scholarship or non-scholarship (Walk-on) athlete (Scholarship: n = 10; Walk-on: n = 4) and were active squad members of the University of Oklahoma’s Men’s Basketball team. This research was approved by the Institutional Review Board of the University of Oklahoma and all participants provided written, informed consent before participating in the study.

In a prospective observational study design, eTL was measured during 22 basketball practice sessions over the course of a 5-week preseason training phase. In addition, weekly measurements of neuromuscular performance were assessed using the CMJ, performed just prior to the start of each strength training session. Subjects performed 1 CMJ assessment prior to the start of the preseason (Pre) and then 1 CMJ assessment per week, following a day off from training, except for Weeks 2 and 3, where CMJ testing was performed in 2 separate sessions due to the logistics of strength training scheduling requiring a portion of the team to come 2 days after an off day. A detailed schedule is provided in Figure 1.

External Training Load (eTL) Monitoring: Subjects wore the Catapult Sport OptimEye T6 IMU system (Catapult Innovations, Melborne, VIC, Australia) comprised of a triaxial accelerometer, gyroscope, and magnetometer, sampling at a rate of 100Hz, in a supportive harness positioned between the scapulae. Subjects wore the same IMU and supportive garment during each practice(McLean et al., 2018) and eTL monitoring started when athletes took the floor for pre-practice warm-ups and ended when they left the floor at the conclusion of practice. Once starting practice, all players remained in the respective drill and were not ‘interchanged’ or substituted, even if they were not the primary participant in the drill as previous literature has suggested substitution can artificially inflate training load intensities (Fox et al., 2018).

All data were analyzed via the Catapult Sport software (Openfield, Catapult, Innovations, Melborne, VIC, Australia) which applies specific algorithms to transform the input of raw inertial data captured during athlete movement into meaningful and standardized output variables used to quantitate the movement experienced. These variables can be classified into two types of variables, “workload variables” and “event detection variables” (Holme, 2015). The workload variables included PL, PL^2D which only includes the accelerometer data from the mediolateral and anteroposterior planes of movement, and individual PL^1D accumulated in the anteroposterior (PL^1D-FWD), mediolateral (PL^1D-SIDE), and vertical (PL^1D-UP) planes of movement. PL/min divides the PL accumulated by time, providing an intensity index.

Inertial Movement Analysis™ (IMA™) values defined as an instant one-step movement effort or micro-movement and is expressed as count data (ct) (Holme, 2015; Spangler et al., 2018; Ward et al., 2018). These distinct acceleration micro-movement events occur during sudden explosive movements, such as accelerations, decelerations, and CoD movements, common among team sport play (Holme, 2015; Spangler et al., 2018; Ward et al., 2018). IMA™ events are detected using proprietary algorithms within the manufacturer software during post-session data analysis generated. IMA™ events are quantified by coupling triaxial accelerometer and triaxial gyroscope data to form a non-gravitational vector and use advanced Kalman filtering algorithms to detect and quantify the frequency of micro-movements experienced during sport play (Holme, 2015). An IMA™ event is detected with the application of polynomial smoothing curves between the start and end point of the accelerative events (Holme, 2015; Spangler et al., 2018). The magnitude of an event (IMA™ Magnitude) is subsequently computed by summing the accelerations under the polynomial curve, measured in terms of delta-velocity, a unit of impulse (m·s^-1).

The key IMA™ variables were as follows: IMA_Low = Low Intensity (1.5-2.5m·s^-1) IMA™ events; IMA_Medium = Medium Intensity (2.5-3.5m·s^-1) IMA™ events; and IMA_High = High Intensity (>3.5m·s^-1) IMA™ events. Additionally, the total number of jump events were combined, as previous literature has outlined the limited sensitivity of the IMU to detect differences in jump heights (Spangler et al., 2018), therefore Jumps = Total number of IMA™ Jump events (including High, Medium, and Low Intensities).

Neuromuscular Fatigue and Performance Assessment: In accordance with previously described methods, (Heishman et al., 2018a; 2018b) following a standardized warm-up and prior to the start of a strength training session, participants performed 3 CMJs on the ForceDecks FD4000 Dual Force Platforms hardware (ForceDecks, London, UK), with a sample rate of 1000Hz.

Subjects started in the tall standing position, with feet placed hip width to shoulder width apart and hands akimbo. Participants then dropped into a self-selected countermovement depth followed by a maximal effort CMJ landing in an athletic position on the force platforms. Subjects reset to the starting position after each jump, and the procedure was completed for 3 jumps. If the subject removed their hands from their hips at any point or exhibited excessive knee or hip flexion once airborne, the jump was ruled invalid and repeated.

ForceDecks software (ForceDecks, London, UK) was used to analyze each CMJ, and the variables of interest were: FT:CT (the ratio of flight time to contraction time), RSI_Mod (calculated as jump height divided by contraction time), and JH (computed by the flight time method) (Heishman et al., 2018a; 2018b; 2019). CMJ tests were performed during the same time of day over the course of the preseason, (Heishman et al., 2017) and the average of the 3 CMJs was utilized for analysis.

All data are reported as means ± SD unless stated otherwise. Statistical analyses were performed using SPSS statistical software with an a priori significance level set at p≤0.05. Initially, data normality was confirmed using the descriptive and graphical information supplemented by Shapiro-Wilk test statistic. A 3-way (Week [W] x Position x Academic) Repeated Measures Analyses of Variance with Bonferroni post hoc analysis was used to examine differences in the average eTL per session each week and differences in the CMJ variables. Effects sizes (Cohen’s d) were calculated and interpreted as trivial (0–0.19), small (0.20–0.49), medium (0.50–0.79), and large (≥0.80) (Cohen, 1992).

PL^1D-UP, PL^1D-SIDE, and PL^1D-FWD contributed 43.7 ± 1.8%, 28.7 ± 1.7%, and 27.3 ± 1.5%, respectively, to PL. The results of the average training load per session each week are outlined in Table 1.

There were no significant differences in the average PL, PL^2D, PL^1D-FWD, PL^1D-SIDE, PL^1D-UP per session each week (p > 0.05). As outlined in Figure 2, there was a significant (p < 0.001) main effect for average PL/min per session, with further analyses revealing significant increases during W1 compared to W3 (p = 0.002, d = 0.46), W4 (p = 0.05, d = 0.60), and W5 (p = 0.015, d = 0.72), and significant increases in W2 compared to W5 (p = 0.035, d = 0.66). No significant differences were observed for IMA_Low (p = 0.163), or IMA_High (p = 0.430), but IMA_Medium presented a significant main effect (p = 0.041) for differences across weeks, however further analyses revealed no significant differences between weeks (p > 0.05).

There were no significant Position*Week interactions and no significant differences in the average PL (p = 0.883), PL/min (p = 0.830), PL^2D (p = 0.794), PL^1D-FWD (p = 0.825), PL^1D-SIDE (p = 0.761), PL^1D-UP (p = 0.852), IMA_Low (p = 0.361), IMA_Medium (p = 0.780), IMA_High (p = 0.780), or Jumps (p = 0.692) per session each week between positions (Table 2). However, a medium effect (d = 0.51) for differences in IMA_Low and a small effect (d = 0.22) for differences in Jumps between positions were observed but the remaining variables displayed trivial effects (d < 0.2).

Significant Academic*Week interactions were observed for PL (p = 0.042), PL^1D-SIDE (p = 0.020), and PL^1D-UP (p = 0.036). Further analyses revealed significantly greater average PL per session for Scholarship athletes during W1 (Scholarships = 841.9 ± 46.0, Walk-ons = 544.4 ± 72.8; p = 0.005), W3 (Scholarships = 792.4 ± 46.5, Walk-ons = 447.6 ± 73.5; p = 0.002), W4 (Scholarships = 721.5 ± 39.3, Walk-ons = 548 ± 62.2; p = 0.036), and W5 (Scholarships = 759.8 ± 23.6, Walk-ons = 526.2 ± 37.3; p < 0.001). Scholarship athletes experienced significantly greater PL^1D-SIDE per session during W1 (Scholarships = 342.7 ± 20.1, Walk-ons = 218.8 ± 31.8; p = 0.007), W3 (Scholarships = 321.64 ± 20.2, Walk-ons = 180.2 ± 32.0; p = 0.002), and W5 (Scholarships = 308.5 ± 12.23, Walk-ons = 206.4 ± 19.3; p = 0.001). Scholarship athletes also experienced greater PL^1D-UP per session during W1 (Scholarships = 545.5 ± 29.5, Walk-ons = 351.6 ± 46.6; p = 0.004), W2 (Scholarships = 499.3 ± 30.0, Walk-ons = 375.8 ± 47.5; p = 0.048), W3 (Scholarships = 514.3 ± 30.5, Walk-ons = 285.5 ± 48.3; p = 0.002), and W5 (Scholarships = 496.0 ± 15.4, Walk-ons = 339.2 ± 24.3; p < 0.001).

As outlined in Figure 3 and Figure 4, there was a significant main effect for differences between Scholarship and Walk-on athletes per session for: PL (p = 0.003, d = 2.2), PL/min (p = 0.002, d = 2.3), PL^2D (p = 0.005, d =2.0), PL^1D-FWD (p = 0.005, d =2.0), PL^1D-SIDE (p = 0.007, d = 1.9), PL^1D-UP (p = 0.003, d = 2.2), IMA_Low (p < 0.001, d = 2.8), IMA_Medium (p = 0.003, d = 2.2), IMA_High (p = 0.003, d = 1.9), and Jumps (p = 0.001, d = 2.6).

There were no significant Position*Week or Academic*Week interactions among any of the neuromuscular performance parameters. As outlined in Figure 5, there were no significant differences in FT:CT, RSI_Mod, or JH across the 5 weeks of preseason. Interestingly, decreases in FT:CT demonstrated a small effect (d = 0.25) from Pre- to W1, while the increase in FT:CT from W2-W3 and W3-W4 showed medium effects (d = 0.55 and 0.61, respectively). Similarly, changes in RSI_Mod demonstrated small effects from Pre- to W1 (d = 0.42), W2-W3 (d = 0.42), W3-W4 (d = 0.35), and W4-W5 (d = 0.46).

There was a significant main effect for differences between positions for JH (Forward/Center = 34.6 ± 0.36, Guards = 42.6 ± 0.36, p < 0.001), but no significant positional differences in FT:CT (Forward/Center = 0.685 ± 0.13, Guards = 0.72 ± 0.13, p = 0.495) or RSI_Mod (Forward/Center = 0.45 ± 0.13, Guards = 0.52 ± 0.13, p = 0.092).

The present study was designed to 1) characterize the average eTL per session across each week of the preseason; 2) examine differences in the average eTL; 3) evaluate the influence of position and academic status on eTL; and 4) examine changes in neuromuscular performance across the 5 weeks of preseason. The main findings of the present study were 1) the characterization of eTL in NCAA Division I basketball players during the preseason training phase; 2) no significant positional differences were observed in eTL; 3) PL^1D-UP contributed less to total PL accumulation during basketball play, while PL^1D-FWD and PL^1D-SIDE each demonstrated a 2-7% greater contribution to total PL compared to the previously reported data on linear running; 4) significant differences in eTL between key players and role players during the preseason training period; and 5) there appeared to be a moderate effect of decreases in FT:CT and RSI_Mod with increases in PlayerLoad/min, however jump height remained unchanged.

This is the first study to provide a comprehensive eTL profile for collegiate men’s basketball players during the preseason training period, which includes the commonly reported PL and PL/min values, but also including IMA™ data. The present study observed greater values for PL than previously reported in collegiate basketball athletes during the preseason (Heishman et al., 2017; 2018a; 2018b) and professional athletes, (Svilar et al., 2018a; 2018b) but more similar to PL and PL/min values seen in semi-professional (Fox et al., 2018). While limited data exists surrounding the IMA™ events in basketball play, the total jumps per session in the present study were approximately 2-3 times greater in addition to total IMA™ events nearly 2 times higher than those previously reported (Svilar et al., 2018a; 2018b). Differences observed in eTL characteristics likely related to the types of drills and activities included in the basketball sessions, which is largely dependent on the sport coaches and their perspective on the needs of the technical and tactical improvements of team, as well as other factors such as the style of play of the team. In addition, it should be mentioned that there is possibility for differences to arise through the use of different IMU hardware throughout the literature. In addition to the use of different manufactures, there is always potential for differences within manufacturer products, with previous much of literature using Catapult Innovations S5 units, while the present study utilized Catapult Innovation T6 devices. Although utilizing the same hardware, a comparison between the two units has yet to be published. All of these factors make the comparison of absolute eTL values between research challenging.

There were no significant differences in eTL variables per week of practice across the preseason, except for increases in PL/min during the first 2 weeks of the preseason. The significant elevation in intensity likely reflects the players’ and coach’s excitement for the new training phase and upcoming season, as well as the later stages of the training phase leading into competition where practices included more instructional time learning tactical strategies, often occurring at lower intensities.

The present study provides evidence that PL^1D-UP contributes 5-15% less to total PL accumulation during basketball activity, compared to laboratory-based studies of linear running activities (Barrett et al., 2014). PL^1D-FWD and PL^1D-SIDE each demonstrated a 2-7% greater contribution to total PL compared to the previously reported running data on linear running (Barrett et al., 2014). These findings are likely due to the large lateral component and intermittent play of basketball requiring frequent accelerations and decelerations (Stojanović et al., 2018). These movements produce more horizontal and fewer vertical ground reaction forces than top-end speed running (Nagahara et al., 2018) that may be achieved in outdoor sports with large areas of play (Cormack et al., 2013; Ward et al., 2018). These findings may suggest the vertical component of basketball activity plays a smaller role in masking minor increases in the mediolateral and anteroposterior vector activity than that experienced in other sports (McLean et al., 2018).

Although positional differences in training load is evident in a variety of sports and are thought to be crucial for improving the individualization of the training program (Aughey and Varley, 2013; Ward et al., 2018), the present study observed no positional differences for eTL (Staunton et al., 2018). Similarly, Staunton et al. (2018) reported few position-specific differences in the exercise dose, average intensity, or the proportion of time spent in each intensity zone during training between front-court and backcourt players within a women’s professional team. Alternatively, previous literature has demonstrated higher acceleration loads for guards, which may be due to smaller players having a lower body mass that can be accelerated with less applied force and tactical principles of the game position bigger players closer to the basket and incorporate a smaller playing zone for their actions, while smaller players travel more distance during gameplay (Schelling and Torres, 2016). Additionally, Svilar et al. (2018a; 2018b) reported key eTL variables were position-dependent; however, their statistical approach limits the generalizability of their findings, and may reflect more to the style of play within their team. Therefore, the lack of difference observed in the present study may be attributed to the team’s style of play or player profiles. Specifically, the present study included several hybrids (or “stretch four”) players that are versatile enough to play around the basket and the perimeter. Further, training phase may contribute to the lack of differences between positions, as the preseason practice strategy differs from that during the season.

These data present a novel finding revealing significant increases, with large effects, in eTL among Scholarship athletes, and ultimately suggests that coaches may need to consider academic differences and player roles when managing eTL. This observation is important as eTL differences may require supplemental training to maintain physical fitness, and be useful for managing training loads, or guiding return-to-play and return-to-performance protocols. The current stratification approach (academics) was used since the player’s roles had not been established during the preseason and misclassification would likely compromise the analysis. Therefore, future research may stratify players considering game-minutes played or team role. Additionally, future literature should explore the eTL disparities during the season since it can be speculated that Walk-on eTL may increase during the competitive phase, as they often relieve scholarship/key players to manage fatigue and may have increased training loads from scout team responsibilities.

To our knowledge, this is the first study to couple quantitative eTL with changes in neuromuscular performance among basketball players. The present study provides conflicting results compared to previous observations reporting no significant differences in JH across the preseason, whereas previous literature has reported increases (Aoki et al., 2017) and decreases (Heishman et al., 2017; Cruz et al., 2018). Similarly, Ferloi et al. (2018) observed no change in JH over the course of the preseason, thus the maintenance of JH across the preseason may suggest that the athletes were able to adequately recover from the prescribed eTL. Therefore, different observations regarding changes in JH over intensive training periods may relate the eTL imposed, but also may be associated with the athlete’s physical capacities and fitness level to with-stand the imposed volumes and intensities of training.

Although there were no significant differences in FT:CT or RSI_Mod observed over the duration of the preseason in the present study, there was a moderate effect for decreases in FT:CT and RSI_Mod following the first 2 weeks of training, which paralleled a significantly greater PL/min. Therefore, JH may not be sensitive enough to detect alterations in eTL parameters, but changes in movement strategy (FT:CT or RSI_Mod) appear to be altered to achieve the gross jump output of JH following intensified bouts of training. These findings support the utility of FT:CT outlined in previous work that has acutely observed substantial reductions in FT:CT following both training and game play (Cormack et al., 2008; Rowell et al., 2017).

The present study has limitations that warrant discussion. First, there was no control of individual work outside of organized team practice, which could influence total eTL and fatigue. Secondly, all players did not jump on the same day due to the logistical challenges when scheduling training session in the collegiate setting (e.g., academic class schedules), however this could have influenced the interpretation of the results. Finally, it may be more practically useful to practitioners if athletes were stratified into groups of key players, rotational players, and developmental players. Due to the training phase the researchers could not definitively delineate each players role in the forthcoming season, however this should be a consideration in future work.

While there were significant increases in practice intensity during W1 and W2 of the preseason, no significant differences manifested in neuromuscular performance indices, suggesting athletes were able to cope with the prescribed training loads. However, there appeared to be small perturbations in FT:CT and RSI_Mod following the weeks with intensified training. Scholarship athletes displayed significantly greater eTL variables when compared to Walk-on athletes, but eTL was not affected by position.

The present study provides valuable insight for performance practitioners, characterizing the eTL profile during practice in a cohort of NCAA Division I basketball players. The significant increases in eTL parameters for Scholarship athletes suggests coaches may need to monitor and manage these athletes with different strategies to maintain fitness and for guiding return-to-play and return-to-performance protocols following injury. Additionally, the lack of positional differences suggests that coaches may not need to stratify teams by position in load monitoring strategies, however coaches should examine the differences among their squad individually, as they may vary depending upon style of play and player personnel. Finally, the present study suggests JH may lack the sensitivity to detect alterations in eTL, while FT:CT and RSI_Mod may be more useful in monitoring neuromuscular performance, as athletes may modify their movement strategy to achieve a desired JH.