Abstracts Track 2026

Area 1 - AI, Data & Sport Systems
Area 2 - Health, Sports Medicine & Rehabilitation Technologies
Area 3 - Performance, Biomechanics & Coaching Support

Area 1 - AI, Data & Sport Systems

Nr: 17
Title:

Internal Load Distribution Across Preseason, In-Season, and Postseason in NCAA Women’s Soccer
Authors:

Victor Romano, Jennifer Austin McCrae and Duncan S Romano
Abstract: BACKGROUND: Understanding how internal training load is distributed across the competitive year is fundamental to applied kinesiology, exercise physiology, and athlete health. However, longitudinal evidence describing phase-specific internal load patterns in women’s collegiate soccer remains limited. PURPOSE: To examine the distribution and magnitude of heart rate–derived internal training load across preseason, in-season, and postseason phases in NCAA women’s soccer athletes. METHODS: A retrospective longitudinal design was used to analyze internal load data from two competitive seasons (2023–2024) in collegiate women’s soccer athletes. Internal load was quantified using Training Impulse (TRIMP), average heart rate expressed as a percentage of maximal heart rate (%HRmax), and time spent above 85% HRmax collected via FirstBeat heart rate monitoring. Sessions were categorized into preseason, in-season, and postseason based on the competitive calendar. Phase differences were assessed using linear mixed effects models with athlete as a random effect, and effect sizes (Cohen’s d) were calculated to determine practical significance. RESULTS: Preseason exhibited the greatest internal load, with mean session TRIMP (~320 AU) significantly higher than in-season values (~300 AU, p < 0.05; d = 0.45) and postseason (~255 AU, p < 0.001; d = 0.93). Average cardiovascular intensity was also highest during preseason (≈72–74% HRmax), with greater time spent >85% HRmax compared with in season (d = 0.41) and postseason (d = 0.78). In season internal load demonstrated reduced variability and relative stabilization despite competitive demands, while postseason load showed marked reductions across all HR derived metrics, consistent with tapering and recovery strategies. No excessive internal load accumulation was observed during postseason phases. CONCLUSION: Internal training load in NCAA women’s soccer is systematically periodized across the competitive year, with the highest cardiovascular stress during preseason, stabilized loading during in-season competition, and deliberate reductions during postseason recovery. PRACTICAL APPLICATION: Phase-specific heart rate monitoring can be used by coaches and sport scientists to objectively verify intended periodization strategies, optimize training stress, and support recovery and long-term athlete development in collegiate team sports.
Nr: 78
Title:

Comparative Analysis of Machine Learning Classifiers for Evaluating Volleyball Dig Positions without Motion Capture Device
Authors:

Jin Kusano and Toshiya Arakawa
Abstract: In sports analytics, optimizing defensive positioning remains a critical challenge. While prior volleyball research has heavily focused on serve reception, objective analysis of dig positioning is still underdeveloped. We analyzed 67 quick attacks. At spike impact, markerless motion capture provided receiver coordinates, which were labeled as successful (get; n=16) or unsuccessful (lost; n=51). LR, SVM, and RF were used to generate spatial probability maps. Results are shown for the Center Back position. The Center Back probability maps did not identify a single optimal coordinate. However, SVM and RF produced distinct spatial probability gradients, suggesting directional tendencies that may inform defensive-position adjustments. Although the current findings suggest directional rather than precise adjustments, this approach provides a basis for future data-driven analysis of defensive positioning in volleyball.

Area 2 - Health, Sports Medicine & Rehabilitation Technologies

Nr: 53
Title:

A Numerical Biomechanical Framework to Assess Tauopathy Onset and Progression for Female Boxing Athletes
Authors:

Jose Gonzales Cabrero, Carlos Gabriel Cardoso, Ricardo Jose Alves de Sousa and Nina Diana Costa
Abstract: 1. Purpose: In this work, a multiscale numerical framework is employed to asess the likeliness of tauopathies onset and spread after repetitive subconcussive events 2. Design/Methodology/Approach: Wearable mouthguards (prevent biometrics) were used to track sparring and combat data on several female boxing athletes. On the numerical side a multi-scale numerical framework encompasses a Finite Element Head developed on Abaqus and a Python code to account for tauopathy progression on space and time. 3. Findings: The first findings of this exploratory work point to the potential of the developed methodology to predict short to long term consequences of sub-concussive events to athletes. 4. Research Limitations/Implications: Several limitations shall be pointed out: The temporal calibration of tau burden was performed using only four post-injury points per region, based on mice data, with a maximum follow-up of 6 months. This is sufficient for a proof-of-concept but needs to be severely improved. 5. Social/Technical Impact: Once developed and validated this tool can definetely contribute as a tool for coaches, doctors and athletes to prepare rehab protocols or stoppage criteria. 6. Final Comments/Originality/Value: This interdisciplinary study can be seen as extremely ambitious but innovative and with great potential for further developments.
Nr: 91
Title:

Transparent Signal Processing Versus Artificial Intelligence for Breathing Rate Monitoring During Sports Activities
Authors:

Jevgenijs Telicko
Abstract: The purpose of this work is to compare two approaches for respiratory monitoring during sports activities: a transparent mathematical signal-processing method and an artificial intelligence-based method for breathing sound analysis. Respiratory rate is an important physiological indicator that can help assess exercise intensity, fatigue, and the athlete’s condition during training. The mathematical approach processes breathing audio as a spectrogram, where inhalation and exhalation phases can be identified in different frequency ranges. The signal is then analyzed using a sliding window, normalization, and peak detection to determine breathing phases and calculate respiratory rate. The main advantage of this method is transparency: each processing step is clear, adjustable, and easy to validate. Therefore, it is especially suitable for laboratory conditions and controlled experiments. However, it is sensitive to background noise, microphone position, and changes in breathing intensity, which may cause false peak detection and require additional filtering. In contrast, the AI-based approach can potentially extract more information from the same respiratory audio signal and may be more precise in real sports conditions, where noise, motion, and individual breathing patterns make fixed mathematical rules less reliable. AI can learn complex patterns from spectrograms and adapt better to variable environments. However, its main limitation is low transparency, as neural networks often work as “black boxes,” making their decisions difficult to interpret and validate in real time. Thus, the mathematical approach provides better explainability and trust, while the AI approach offers greater potential accuracy and richer information extraction. Results show that, under specific conditions, both methods can achieve high accuracy, with an error below 10%. Therefore, both approaches may be suitable for specific tasks.
Nr: 92
Title:

Combined Effects of Acute Normobaric Hypoxia and Partial Sleep Restriction on Physical and Cognitive Performance in Physically Active Young Men: A Randomized Crossover Pilot Study
Authors:

Mateusz Świątek, Zuzanna Komarek, Maria Lipowska, Jacopo Vitale, Małgorzata Chołda, Magdalena Więcek and Aleksandra Filip-Stachnik
Abstract: Acute hypoxia and sleep restriction commonly occur during mountain expeditions and may impair physical and cognitive performance. Their combined effects remain insufficiently understood. This study investigated the independent and interactive effects of simulated hypoxia and partial sleep restriction. A randomized crossover study was conducted in five healthy physically active men (21.2 ± 1.9 years). Participants completed four randomized, counterbalanced conditions: hypoxia or normoxia after habitual sleep or 3-h sleep restriction. During sleep restriction, participants woke up 3 h earlier than usual to simulate sleep curtailment commonly experienced during mountain expeditions. Sessions were separated by at least 72 hours and were performed in a normobaric chamber (FiO₂ ≈ 15%; ~2500m). Following three consecutive 45-min low-intensity cycling bouts, cognitive performance (Flanker Task), handgrip strength endurance, countermovement jump, pull-up endurance, and perceived exertion were assessed. Two-way ANOVA revealed significant main effects of partial sleep restriction with reduced countermovement jump maximal power (p = 0.044, η²p = 0.678) and pull-up endurance (p = 0.016, η²p = 0.800). No significant main effects of hypoxia or hypoxia × sleep interactions were observed for any outcome. This pilot study provides preliminary evidence that partial sleep restriction may adversely affect selected physical performance outcomes, whereas no significant effects of acute moderate hypoxia were observed in this pilot sample. These findings highlight the importance of adequate sleep before altitude exposure and warrant confirmation in larger studies. These findings may inform evidence-based recommendations for sleep management before altitude exposure to maintain physical performance in hypoxia. This pilot study provides novel preliminary data on the combined effects of acute normobaric hypoxia and partial sleep restriction, supporting future larger-scale investigations.

Area 3 - Performance, Biomechanics & Coaching Support

Nr: 82
Title:

Modelling Set Outcomes in Elite Volleyball through Rotation Matchup Analysis and Monte Carlo Simulation
Authors:

Sotirios Drikos, Sotirios Drikos and Maxence Goualou
Abstract: Purpose. At the elite volleyball level, sets are often decided by 2–3 points. Standard analysis relies on aggregate team metrics, ignoring performance variances across rotational configurations. This study proposes a stochastic theoretical framework using empirical rotational probabilities and Monte Carlo simulations to model set-level win probabilities, providing coaches with a decision-support methodology to optimize starting rotations and assess tactical risk. Methodology. The study combines empirical win-rate estimation with a stochastic Monte Carlo model. Match files from the 2026 VNL preliminary round were merged with FIVB roster data scraped via the R package rvest to track setter positions (1, 6, 5, 4, 3, 2). A 6 X 2 Bernoulli success probability matrix PK was built per team for serve and reception contexts. The rally win probability balances Team A's serving strength and Team B's reception weakness. Point-by-point simulations incorporate scoring rules, rotation transitions, and consecutive-serve constraints. For all 72 starting configurations, N=10,000 iterations guaranteed precision within ±0.5%. Findings. Applied to the 2026 VNL Finals (180,000 iterations per matchup), initial rotation alignments yielded win probability shifts exceeding 5–6% in close contests. An interactive R Shiny dashboard (VNL Rotation Analyser) was deployed for real-time probability matrix visualization and rotation optimization. Research Limitations/Implications Calibrated on pool-stage data, the model omits home-court advantage, schedule imbalance adjustments, and final-round intensity surges. Technical Impact Embeds complex simulations behind a user-friendly interface, enabling staff to convert raw scouting data into actionable, risk-quantified tactical strategies. Value Evaluates set outcomes through explicit rotation interactions, providing a scalable forecasting tool for high-performance sports environments.
Nr: 16
Title:

Incorporating Haptics to Maximize Performance, Prevent Injury, and Assist Recovery in Elite Athletes
Authors:

Peter Hurwitz, Jason Taporco and Mark Sakr
Abstract: Optimizing athletic performance relies on ensuring proper muscle tone, strength, flexibility, and movement coordination. Neuromuscular training to improve athletic performance has been embraced in athletic programs to aid athletes in maximizing their potential in national/international athletic competitions. The POWER-UP Randomized Controlled Trial accessed whether haptic patch technology could improve athletic muscular strength, power, and endurance. Methods Healthy Division I college athletes, (N=70; 30 males and 40 females), were enrolled in this blinded case-control study. While 34 athletes received a patch (VICTORY patch) with haptic vibrotactile trigger technology (VTT), 36 athletes received a placebo patch. Neuromuscular performance evaluation methods and validated scales were incorporated. Outcomes included neuromuscular changes in specified muscle groups, output jumping and speed testing, and measurements pertaining to complex neuromuscular pattern performance. Neuromuscular data was collected prior to first-patch use; at 2-3 days, 7 days, and 9-10 days post first-patch use, respectively. Descriptive statistical methods inclusive of Student’s t-test and ANOVA testing were used for data analysis. Results Comparative results including peak power, concentric peak force, jump height, and knee extension/flexion were obtained and data analysis compared results between the active and placebo patch groups. Active patch subjects showed statistically significant improvements at 9-10 days post first-patch use, and greater improvement overall in the evaluated neuromuscular attributes compared to the placebo patch group. Additionally, non-dominant leg 'active' patch placement resulted in the improvement of non-dominant leg performance outcomes and be in stasis with the dominant leg as compared to no significant change in the placebo patch group. Conclusions Incorporating haptic patch use in athletic training programs to improve neuromuscular functioning may be beneficial to improve and maximize athletic performance, injury prevention, or to recover lost strength, power, or endurance.
Nr: 29
Title:

Beyond Data Collection: A Translational Framework for Wearables, Artificial Intelligence and Coach-Oriented Decision-Making in Sport
Authors:

Pedro Morouço and Silvia Fantozzi
Abstract: Wearable technologies, inertial measurement units, force sensors, computer vision and artificial intelligence are increasingly embedded in sport science and coaching environments. However, the growing availability of biomechanical and physiological data has not always translated into better training decisions, improved athlete monitoring or more effective performance support. A persistent gap remains between data acquisition, data interpretation and coach-oriented action. This position paper argues that the next stage of technological development in sport should move beyond sensor validation and data accumulation towards translational, decision-support frameworks that are understandable, context-sensitive and operationally useful for coaches, athletes and multidisciplinary teams. Drawing on examples from biomechanical analysis, swimming performance, wearable monitoring and artificial intelligence, the paper proposes a conceptual pathway linking data capture, signal processing, feature extraction, pattern recognition, contextual interpretation and actionable coaching feedback. The central argument is that sport technologies should not be evaluated only by their technical accuracy, but also by their capacity to support timely, meaningful and individualized decisions in real training and competition settings. Three priorities are discussed: first, the need to align technological outputs with sport-specific performance determinants; second, the importance of explainable and coach-friendly artificial intelligence models; and third, the requirement for ecological validation in applied environments. The paper concludes by proposing a translational framework for intelligent coaching support systems, emphasizing interdisciplinary collaboration between sport scientists, engineers, data scientists, coaches and clinicians. Such a framework may help ensure that emerging technologies enhance, rather than complicate, the decision-making process in sport.
Nr: 89
Title:

Ground Reaction Forces and Race Walking Economy
Authors:

Tapani Keränen, Esa Hynynen and Jani Lehtinen
Abstract: The purpose of this study was to identify race walk (RW) step contact ground reaction force (GRF) variables related to physiological RW economy at highly trained athletes’ race pace. The data were collected during race walk tests performed on a 200 m indoor track at the High Performance Institute KIHU in Jyväskylä, Finland, between 2018 and 2020. During this period, the athletes completed the test three to four times per year. Each test consisted of 5–8 × 1000 m RW stages performed at progressively increasing intensities while maintaining RW technique. The study included 10 Finnish national team race walkers (five women and five men), all experienced international competitors. During the tests, oxygen uptake was measured continuously to determine physiological RW economy. In total, 19 vertical and horizontal GRF variables were obtained (1000 Hz) using a 5.73 m long chain of force plates. For each athlete, the most economical RW stage at their individual 20 km (n = 7) or 50 km (n = 3) race pace and a less economical RW stage at a similar velocity were selected from two different test sessions. The analysis was based on within-subject difference variables (uneconomical − economical). Paired-samples t-test and Partial Least Squares Regression (PLS) with Variable Importance in Projection (VIP) scores were used. Economical and less economical RW velocities were 3.50 ± 0.34 and 3.49 ± 0.32 m·s⁻¹, with corresponding oxygen consumptions of 249 ± 14 and 267 ± 14 mLO₂·kg⁻¹·km⁻¹ (p < 0.001), respectively. A PLS model including Δ Max Braking Force, Δ Vertical Max Force Timing, Δ Resultant Force at Propulsion Phase, Δ Average Vertical Force at Propulsion Phase and Δ Average Vertical Force explained 65% of the variance in RW economy (R² = 0.65). Minimizing braking force, optimizing vertical loading at midstance, and maintaining consistent force production during propulsion phase form a coordinated biomechanical force-production strategy associated with RW economy.
Nr: 94
Title:

Introducing a Novel Ball Game into University Lectures: A Performance Analysis Approach for Innovative Teaching.
Authors:

Takuya Magome and Takeshi Tanaka
Abstract: 1. Purpose University physical education requires encouraging exercise among students with diverse backgrounds, yet quantitative evaluation remains scarce. Leveraging sensor-based analytics from competitive sports [1, 2], this study evaluates exercise promotion effects and cooperation dynamics of a novel ball game to establish a quantitative framework for program optimization. 2. Design/Methodology/Approach Conducted over four months with 90 students (analyzing 45 participants), sensor measurements were taken at the beginning and end of the course. Participants wore a 3-axis accelerometer (20 Hz) on their dominant arm to estimate movement intensity, while a laser radar system monitored real-time positional coordinates. Integrated datasets were analyzed to quantify team dynamics. 3. Findings Spatial tracking showed that while players mainly occupied the central area, scoring plays effectively utilized court sides with higher breakthrough acceleration. Crucially, average exercise intensity increased significantly over the lectures, confirmed by a paired t-test (p < .001 ). 4. Research Limitations/Implications Limitations include a sample size of N = 45 at a single institution. However, implications show that combining spatial and motion sensing provides a scalable methodology for longitudinal physical education evaluation. 5. Social/Technical Impact Technically, this work demonstrates non-invasive multi-sensor integration for general education settings. Socially, it empowers instructors with objective metrics and provides students with visual feedback to enhance intrinsic motivation and health-related QOL. 6. Final Comments/Originality/Value This research translates competitive sports analytics into general physical education via a novel ball game, providing a practical, data-driven model for innovative teaching. References: [1] Magome, T., et al., IPSJ Journal, 2021. [2] Magome, T., et al., Proc. 12th icSPORTS, pp. 160-167, 2024.
Nr: 97
Title:

A Virtual Reality-Based System for Evaluating Feed-Forward and Feed-Back Control of Spatiotemporal Swing Adjustment in Baseball Batting: Toward Practical Training Applications
Authors:

Futa Tsuruhara, Toshitaka Kimura, Akemi Kobayashi, Sorato Minami and Maki Sugimoto
Abstract: Baseball batting requires decision-making and swing execution under severe time constraints, demanding feed-forward (FF) and feed-back (FB) control. FF/FB are usually evaluated separately; this study asks whether VR can dissociate and quantify both, and how much each explains performance. Fifteen university players completed three VR conditions: Conditions 1 and 2 each targeted a distinct control process, and Condition 3 yielded performance measures examined against these indices via path analysis. Condition 1 assessed temporal FF/FB adjustment via alternating fast (straight) and slow (change-up) pitches: FF = onset-time diff (delta_onset_time), FB = acceleration diff (delta_acceleration). Condition 2 assessed online spatial FB adjustment using straight and straight-to-slider pitches, the latter shifting trajectory mid-flight—a VR-only stimulus requiring mid-swing correction; SEM (Structural Equation Modeling) flagged deviation (swing_deviation_flag). Condition 3 randomly presented four pitch types (straight, change-up, slider, straight-to-slider) to obtain performance measures (miss-swing rate, batted-ball speed); path analysis then examined their link with the three indices. Results showed good fit (chi-square p>0.05; CFI, TLI=1.00). delta_onset_time and delta_acceleration significantly predicted performance (p=0.0301, p=0.0001); swing_deviation_flag did not (p=0.059). Of 15 participants, 4 were regulars, and K-means placed 3 in the best cluster, suggesting the evaluation reflects real batting ability. A limitation is that some found HMD weight/depth perception uncomfortable; as all were VR-naive, adaptation may have skewed results. Socially, this could yield a new training evaluation index via clustering. Understanding batting ability requires an integrated framework, not separate evaluation; this study's originality lies in quantifying FF and FB in one VR system, jointly linking them to performance. Future work: larger samples, pros, real-world test.