Abstract
aceRunning is a sport for disabled people and successful performance depends on reduc- ing the amount of time spent travelling a specific distance. Performance analysis in RaceRunning athletes is based on traditional methods such as recording race time, distances travelled and fre- quency (sets and reps) that are not sufficient for monitoring training loads. The aims of this study were to monitor training loads in typical training sessions and evaluate technical adaptations in RaceRunning performance by acquiring sensor metrics. Five elite and competitive RaceRunning athletes (18.2±2.3 yrs ) at RR2 and RR3 levels were monitored for 8 weeks, performing in their usual training sessions while wearing unobtrusive motion sensors. The motion sensors were attached to the waist and lower leg in all training sessions, each lasting between 80 and 90 min. Performance metrics data collected from the motion sensors included player loads, race loads, work/rest ratio and impact shock directions, along with training factors (duration, frequency, distance, race time and rest time). Results showed that weekly training loads (player and race loads) followed acceptable threshold levels, according to assessment criteria (smallest worthwhile change, acute/chronic work ratio). The relationship between race velocity (performance index) and race load was non-linear and statistically significant, which led to different performance efficiency groups. Wearable motion sensor metrics revealed small to moderate technical adaptations following repeated sprint attempts in temporal running performance, variability and consistency. In conclusion, using a wearable-based system is an effective feedback
criteria (smallest worthwhile change, acute/chronic work ratio). The relationship between race velocity (performance index) and race load was non-linear and statistically significant, which led to different performance efficiency groups. Wearable motion sensor metrics revealed small to moderate technical adaptations following repeated sprint attempts in temporal running performance, variability and consistency. In conclusion, using a wearable-based system is an effective feedback tool to monitor training quality, revealing important insights into adaptations to training volumes in disabled athletes. Keywords:RaceRunning; adaptations; training loads; wearable-motion sensors; performance efficiency; feedback 1. Introduction The number of disabled people who participate in sports is growing because of the noted positive impact of sports participation on wellbeing, health-related fitness com- ponents and motivation [1]. RaceRunning is a para sport: an adapted form of running for disabled people requiring use of a three-wheeled bike during locomotion. Due to its ergonomic design, it provides an opportunity for disabled people to move, despite severe limb impairments [2]. RaceRunning is a competitive track event for people with severe coordination impairments [3] and was registered as a para sport in 2017, initially for people with brain injuries, later including cerebral palsy (CP) patients [4]. Due to the inclusive nature of this sport for different groups of disabled individuals, it can provide valuable opportunities for increasing the participation of disabled people in organised physical activity for their health and wellbeing and also for competitive purposes. Applications of technology in modern sports have become more common, due to their impact on performance. Nowadays, many team and individual sports benefit from aug- mented informational feedback that is provided by technologies to enhance performance, monitor training volumes, prevent injuries and evaluate the effectiveness of practice on Sensors2024,24, 7923.
Sensors2024,24, 7923 2 of 16 skills and physical fitness [5]. For some time, performance analysis in sport, using technolo- gies such as video analysis systems and wearable sensors, has been an important element of the coaching process [6]. The significant roles of the support technology have been appreciated within the cycle of competition, reflection, decision-making and performance preparation [7] by coaching and sport scientists. The role of technology in practice and competition for people with disabilities is paramount, too. In a systematic review of 39 studies [8], the importance of wearable tech- nologies for monitoring practice and activity in disabled athletes has been emphasised. Results revealed that the two common technologies used for performance assessment were motion sensors and portable electromyography (EMG) systems. The authors categorised their applications into four distinct groups: athlete classification, injury prevention, perfor- mance characterisation/training optimisation and equipment customisation. Whilst the selected sports were mainly wheelchair sports (rugby, basketball, racing and curling), there were three studies in running and one study in RaceRunning and in people with CP that indicated a lack of available evidence in RaceRunning sport. Using wearable systems for assessment of running, jogging and sprinting perfor- mance in both practice and competition is common [9]. For example, wearable systems provide valuable information on physical exertion and movement economy, evaluated according to travelled distance, velocity, acceleration and deceleration profiles, applicable for planning strength and conditioning programmes. The wearable motion sensors that are used frequently in sports are inertial measurement units (IMUs) and global tracking systems (GPS). They are integrated into different hardware components (accelerometer, gyroscope, magnetometer) in a small case, using a conventional metric, player load (a converted form of three-axis acceleration signals), to assess acute (daily time scale) and chronic (aggregated weekly and monthly time scales) workloads. These sensors have been used for monitoring training workload and prediction of overuse injuries in sports such as rugby [10], football [11], volleyball [12] and swimming [13]. Despite the popularity of running in different sports, running-related injuries are still prevalent [14] due to an imbalance between training and recovery [15]. One way to estimate workloads and
monthly time scales) workloads. These sensors have been used for monitoring training workload and prediction of overuse injuries in sports such as rugby [10], football [11], volleyball [12] and swimming [13]. Despite the popularity of running in different sports, running-related injuries are still prevalent [14] due to an imbalance between training and recovery [15]. One way to estimate workloads and predict risk of overuse injuries in runners is by using wearable motion sensors. Reports from different cohorts of runners have revealed that such systems are feasible and informative for guiding training sessions, because of their handy size and the meaningfulness of data analytics platforms to monitor performance over time [16–18]. For example, Cloosterman et al. [17] showed that GPS data were functional in calculating weekly acute-to-chronic workload ratio (ACWR) and associations between training load and onset of running-related knee injuries in recreational runners. ACWR is a useful metric to calculate an athlete’s ability to tolerate sudden changes in load. It also is a valid predictor of risk of overuse injuries in sports. For example, in rugby players, an ACWR value above 2.0 predicted the likelihood of injuries [19]. Neal et al. [16] reported 70% adherence and 92% successful data collection in recreational runners through using a wrist IMU/GPS sensor in monitoring acute training loads for prediction of injury. One concept relevant to sports injuries is training adaptation, which indicates the body’s response to training stress [20]. To achieve optimal individual performance in running, coaches usually manipulate some training factors related to external loads, such as intensity, frequency, duration, distance and number of repetitions or training volume [21]. Adding other training metrics, such as external or internal workload metrics (e.g., cumula- tive shock, rating of perceived exertion), to the conventional training methods of runners could provide valuable information to individualise training adaptations and potentially reduce the risk of overtraining and overuse injuries [22]. The number of studies that have used wearable motion sensors in monitoring training loads in disabled athletes for performance enhancement or injury prevention is limited. Fulton et al. [23] used IMU sensors in monitoring Paralympic swimmers to investigate
methods of runners could provide valuable information to individualise training adaptations and potentially reduce the risk of overtraining and overuse injuries [22]. The number of studies that have used wearable motion sensors in monitoring training loads in disabled athletes for performance enhancement or injury prevention is limited. Fulton et al. [23] used IMU sensors in monitoring Paralympic swimmers to investigate the role of kicking in freestyle swimming, by quantifying variables like kick count, rate and amplitude. To investigate changes in training load according to a specific athlete’s
Sensors2024,24, 7923 3 of 16 activity, some studies have used motion sensors on the wheelchair frame to obtain data on performance parameters such as mean linear acceleration, rotational velocity and acceleration in wheelchair basketball [24] and wheelchair tennis [25] or for computation of energy expenditure and intensity level of players in wheelchair rugby [26]. Furthermore, heart rate sensors have been used to monitor training load in running [27,28] and wheelchair basketball [24]. There is no evidence to suggest that the training workloads in RaceRunning athletes need to be the same as those of able-bodied individuals, except through use of conventional methods of recording indirect training variables (number of races undertaken, time, dis- tance, etc.). This method also is not adequate to gain real-time data on the body’s responses to training stress (volume). Hence, using wearable sensors to collect more information about body impact shock (running loads) could help coaches and trainers to optimise training programmes based on training feedback for enhancing performance and reducing risks of overtraining and overuse injuries, specifically in disabled athletes who often have structural and functional variations to contend with. Thus, the aims of this study were to monitor training loads in typical training sessions, and evaluate technical adaptations in RaceRunning performance from sensor metrics. 2. Methods 2.1. Participants The study used a descriptive/prospective design in which the training status of the participants was recorded without any intervention. Five (two male and three female) elite and competitive athletes (age: 18.2±2.3 years; body mass: 51.21±5.4 kg; and height: 167.1±6.5 cm) were non-randomly selected from a local RaceRunning club. Because of the purpose of the study and its descriptive nature, all members of the club were recruited non-randomly. The eligibility criteria were disabled athletes at levels of RR2 and RR3 RaceRunning levels (RR2: n = 2 and RR3: n = 3), according to CP International Sports and Recreation Association classifications. Athletes in the RR2 class have spasticity, athetosis, ataxia dystonia, or muscle weakness, which limit the effective pushing movements of the lower extremities. Athletes in the RR3 class have mild to moderate involvement in one or both upper
and RR3 RaceRunning levels (RR2: n = 2 and RR3: n = 3), according to CP International Sports and Recreation Association classifications. Athletes in the RR2 class have spasticity, athetosis, ataxia dystonia, or muscle weakness, which limit the effective pushing movements of the lower extremities. Athletes in the RR3 class have mild to moderate involvement in one or both upper extremities, fair to good trunk control, and moderate involvement of the lower extremities. Other eligibility criteria were long-term neurological conditions, including spastic cerebral palsy (n = 4) and acquired brain injuries (n = 1), freedom from any musculoskeletal injury during data collection and participation in competitions (mean experience: 3.0±0.7 years). Their level of ambulation was assessed using the Functional Mobility Scale [29], in which they were assessed on their perceived ability to walk different distances (5 m, 50 m, 500 m) independently (rate = 6) to using a wheelchair (rate = 1). The participants rated their ability at 6 in the 5 m distance and at 1 in the 500 m distance. Participants completed a consent form in the presence of their carers. The study was approved by an institutional University research ethics committee and conducted according to the ethical guidelines of the Helsinki Declaration of 1964. 2.2. Materials The main components of the performance analysis system in this study were 9-axis (3-axis Accelerometer, 3-axis Gyroscope, 3-axis Magnometer) and low-mass (<3 g) wearable motion sensors (MetaMotion R, MBIENT LAB Co., San Jose, CA, USA). The sensors were equipped with Bosch Sensortec (Stuttgart, Germany), which combines measurements of the accelerometer, gyroscope and magnetometer to provide a robust calculation of the orientation vector (3-axis Euler angle). The 2 motion sensors were used throughout the performance analysis period for capturing training loads on different body parts. A waist sensor was attached to the low-back area (L2–L3) for measuring whole body training load, and a leg sensor was attached to the medial-distal part of the right tibia for measuring lower-limb impact shock
different body parts. A waist sensor was attached to the low-back area (L2–L3) for measuring whole body training load, and a leg sensor was attached to the medial-distal part of the right tibia for measuring lower-limb impact shock
Sensors2024,24, 7923 4 of 16 as well as recognising running phases (stance, flight and stride). For detecting running phases, the gyroscope and accelerometer of the tibia sensor were synchronised. The tibia sensor has previously been validated for use in different activities [30]. The sensors were secured by double-sided tape and Velcro adjustable straps (Presco, Swindon, UK). Motion sensor orientation was calibrated by the sensor–body alignment. The tibia sensor was placed so that the X axis was aligned with the shank length in the standing position (X: superior–inferior; Y: anterior–posterior; Z: mediolateral). The waist sensor alignment was 90 degrees rotation relative to the tibia sensor (X: mediolateral; Y: superior–inferior; Z: anterior–posterior). The sensor’s sample rate was set at a frequency of 400 Hz. The motion sensors were programmed by a free mobile application (MetaBase, MBI- ENT LAB, Co., San Jose, CA, USA). MetaBase is a user-friendly application that runs on both iOS and Android platforms. This application can synchronise sensors for simultane- ous data capturing, saving and exporting. In addition, it was possible to customise data collection in terms of signal type (acceleration, gyroscope, etc.), speed (25 Hz to 800 Hz) and transmission mode (streaming, logging). For this study, all sessions were recorded through the logging mode, and raw data were exported as a CSV file for further analysis. A Polar Heart Rate sensor (Polar Sense armband and chest strap) was used to monitor internal load during the training session. The Polar armband is an optical sensor that was wrapped around the right upper arm and connected via Bluetooth to the Polar mobile application (Polar Flow App, version 6.24.0) for recording the heart rate per athlete. 2.3. Procedure The data collection protocol was followed according to Figure. The principal inves- tigator (MS) was a performance analyst in the RaceRunning club who worked with the participants and the coaching team to discuss the protocol. Some stages of the protocol, such as data collection, feedback provision and training monitoring (see Figure), required effective communication with the coach and athletes to enhance the viability and feasibility of the wearable-based system in the field. Data
(MS) was a performance analyst in the RaceRunning club who worked with the participants and the coaching team to discuss the protocol. Some stages of the protocol, such as data collection, feedback provision and training monitoring (see Figure), required effective communication with the coach and athletes to enhance the viability and feasibility of the wearable-based system in the field. Data collection took place at an indoor athletics track where the participants trained for one day per week. Participants wore standard running shoes and clothing, and everyone had to use a RaceRunner bike (Petra Cross Runner, Quest 88 Ltd., Shifnal, UK) which was adjustable in terms of body dimensions. The coach supervised the training session, which consisted of a routine programme including warming up with stretching, low-velocity running and a main part that was planned based on the seasonal training volume in terms of the number of runs, distance and intensity. Sprint running occurred on a straight line track, ranging between 20 m and 100 m in distance. Usually, the sprint running (activity) period was followed by a rest period of 7–8 min for a full recovery. The rest periods were dynamic and included slow walking and active stretching. The standard procedure for monitoring training volumes was the pen-paper method (using a training log notebook) in which the coach wrote the number of runs, their distances and the race time for each participant. The wearable-based performance analysis system was added to the traditional methods in this study, providing an objective assessment tool in sprint performance for assessing training loads and race intensity, and as a monitoring system to individualise optimal loads and prevent any risk of overtraining. The motion sen- sors were attached to the participant’s body before the start of the main training component and were removed after the cooling down period. In addition, the principal investigator recorded the start and the end of the training session to match it with the sensor timestamp (year/month/date/time). Each session lasted between 80 and 90 min, and the length of this study was 2 months. The coaches and athletes were regularly provided group
the main training component and were removed after the cooling down period. In addition, the principal investigator recorded the start and the end of the training session to match it with the sensor timestamp (year/month/date/time). Each session lasted between 80 and 90 min, and the length of this study was 2 months. The coaches and athletes were regularly provided group and individual delayed feed- back (1 week) on the quality of training sessions, based on the defined training metrics/key performance indicators (KPIs) in RaceRunning (see the next section). The type of feedback was mainly provided as visual feedback in Excel charts. Over time, session-by-session
Sensors2024,24, 7923 5 of 16 variations and fluctuations in KPIs were provided in PowerPoint slides to facilitate strategic decision-making in the training plan.Sensors 2024, 24, x FOR PEER REVIEW 4 of 16 area (L2–L3) for measuring whole body training load, and a leg sensor was attached to the medial-distal part of the right tibia for measuring lower-limb impact shock as well as rec- ognising running phases (stance, flight and stride). For detecting running phases, the gy- roscope and accelerometer of the tibia sensor were synchronised. The tibia sensor has pre- viously been validated for use in different activities [30]. The sensors were secured by double-sided tape and Velcro adjustable straps (Presco, Swindon, UK). Motion sensor ori- entation was calibrated by the sensor–body alignment. The tibia sensor was placed so that the X axis was aligned with the shank length in the standing position (X: superior–inferior; Y: anterior–posterior; Z: mediolateral). The waist sensor alignment was 90 degrees rota- tion relative to the tibia sensor (X: mediolateral; Y: superior–inferior; Z: anterior–poste- rior). The sensor’s sample rate was set at a frequency of 400 Hz. The motion sensors were programmed by a free mobile application (MetaBase, MBI- ENT LAB, Co., San Jose, CA, USA). MetaBase is a user-friendly application that runs on both iOS and Android platforms. This application can synchronise sensors for simultane- ous data capturing, saving and exporting. In addition, it was possible to customise data collection in terms of signal type (acceleration, gyroscope, etc.), speed (25 Hz to 800 Hz) and transmission mode (streaming, logging). For this study, all sessions were recorded through the logging mode, and raw data were exported as a CSV file for further analysis. A Polar Heart Rate sensor (Polar Sense armband and chest strap) was used to monitor internal load during the training session. The Polar armband is an optical sensor that was wrapped around the right upper arm and connected via Bluetooth to the Polar mobile application (Polar Flow App, version 6.24.0) for recording the heart rate per athlete. 2.3. Procedure The data collection protocol was followed according to Figure 1. The principal inves-
Description
This study monitors training loads and evaluates technical adaptations in RaceRunning performance using wearable sensors.