Stefan Kranzinger
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Stefan Kranzinger has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Christina Kranzinger
Salzburg Research Forschungsgesellschaft mbH, Salzburg, Austria
7 shared publications - Thomas Stöggl
Department of Sport and Exercise Science University of Salzburg Salzburg Austria
5 shared publications - Severin Bernhart
Salzburg Research Forschungsgesellschaft mbH, Salzburg, Austria
4 shared publications - Karl Rehrl
Salzburg Research Forschungsgesellschaft mbH, Austria
3 shared publications - Aleksandar Tomaskovic
Department of Sports Medicine, Rehabilitation and Disease Prevention, University of Mainz, Mainz, Germany
2 shared publications Christina Kranzinger
Human Motion Analytics, Salzburg Research Forschungsgesellschaft mbH, Salzburg, Austria
2 shared publications
Latest Publications
- Open Access
- 12 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 213534-213546
User Acceptance and Behavioral Impact of a Wearable Ski Sensor System
The integration of wearable sensor technology into recreational alpine skiing has the potential to improve technique, engagement, and safety through real-time feedback. This study evaluates the usability and perceived benefits of the Connected Boot (CTD-Boot), a prototype system that assesses ski performance via cuff-mounted inertial measurement units and a mobile phone application. In a living-lab design with independent skiing at user-chosen resorts, data from 89 recreational skiers were collected. Using an adapted Technology Acceptance Model, we examined user acceptance, predictors of behavioral change, and the system’s potential for skill development. A Random Forest classifier predicting self-reported improvement achieved 60.9% accuracy with fair agreement (Cohen’s = 0.378), and identified “perceived usefulness” and “ease of learning” as the strongest predictors. Qualitative feedback underscored the value of personalized, real-time insights. Overall, wearable systems such as the CTD-Boot can support skill enhancement when interaction is intuitive and information is actionable and enjoyable, offering clear implications for the design of digital coaching tools in skiing.
- 12 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 213534-213546
Open AccessUser Acceptance and Behavioral Impact of a Wearable Ski Sensor System
The integration of wearable sensor technology into recreational alpine skiing has the potential to improve technique, engagement, and safety through real-time feedback. This study evaluates the usability and perceived benefits of the Connected Boot (CTD-Boot), a prototype system that assesses ski performance via cuff-mounted inertial measurement units and a mobile phone application. In a living-lab design with independent skiing at user-chosen resorts, data from 89 recreational skiers were collected. Using an adapted Technology Acceptance Model, we examined user acceptance, predictors of behavioral change, and the system’s potential for skill development. A Random Forest classifier predicting self-reported improvement achieved 60.9% accuracy with fair agreement (Cohen’s = 0.378), and identified “perceived usefulness” and “ease of learning” as the strongest predictors. Qualitative feedback underscored the value of personalized, real-time insights. Overall, wearable systems such as the CTD-Boot can support skill enhancement when interaction is intuitive and information is actionable and enjoyable, offering clear implications for the design of digital coaching tools in skiing.
- Open Access
- 29 November 2025
- Published by Springer Nature in Discover Artificial Intelligence
- Vol. 6 (1)
A scoping review of explainable artificial intelligence in sports science
Artificial intelligence (AI) and machine learning (ML) are increasingly being applied in sports science to predict performance, assess injury risk, and support tactical decision-making. However, the opacity of many machine learning (ML) models has led to a growing interest in explainable AI (XAI) methods, which aim to make the decisions of "black box" algorithms more transparent and interpretable. The result of this is an increase in trust, facilitation of model validation, and enablement of more informed decision-making for practitioners and analysts. While XAI is well established in domains like healthcare or finance, its application in sports science remains fragmented and underexplored. This scoping review follows the PRISMA-Sc approach and systematically examines how XAI and interpretable ML techniques have been applied in the context of sports science. Between 2014 and June 2024, a total of 19 studies were identified through a multi-database search and analyzed in terms of sport type, ML model, XAI method, dependent variable, limitations, challenges and future research. The results reveal a clear dominance of SHapley Additive Explanations across multiple disciplines, while other tools such as Gradient-weighted Class Activation Mapping, Individual Conditional Expectation Plots, or domain-specific approaches are rarely used. Visual and rule-based explanation methods were largely absent. The findings highlight a lack of methodological diversity and limited validation of explanations with domain experts or practitioners. To increase real-world impact, future work should explore comparative evaluations of XAI techniques, domain-specific explanation frameworks, and user-centered approaches that align with the needs of coaches, athletes, and analysts.
- 29 November 2025
- Published by Springer Nature in Discover Artificial Intelligence
- Vol. 6 (1)
Open AccessA scoping review of explainable artificial intelligence in sports science
Artificial intelligence (AI) and machine learning (ML) are increasingly being applied in sports science to predict performance, assess injury risk, and support tactical decision-making. However, the opacity of many machine learning (ML) models has led to a growing interest in explainable AI (XAI) methods, which aim to make the decisions of "black box" algorithms more transparent and interpretable. The result of this is an increase in trust, facilitation of model validation, and enablement of more informed decision-making for practitioners and analysts. While XAI is well established in domains like healthcare or finance, its application in sports science remains fragmented and underexplored. This scoping review follows the PRISMA-Sc approach and systematically examines how XAI and interpretable ML techniques have been applied in the context of sports science. Between 2014 and June 2024, a total of 19 studies were identified through a multi-database search and analyzed in terms of sport type, ML model, XAI method, dependent variable, limitations, challenges and future research. The results reveal a clear dominance of SHapley Additive Explanations across multiple disciplines, while other tools such as Gradient-weighted Class Activation Mapping, Individual Conditional Expectation Plots, or domain-specific approaches are rarely used. Visual and rule-based explanation methods were largely absent. The findings highlight a lack of methodological diversity and limited validation of explanations with domain experts or practitioners. To increase real-world impact, future work should explore comparative evaluations of XAI techniques, domain-specific explanation frameworks, and user-centered approaches that align with the needs of coaches, athletes, and analysts.
- Open Access
- 30 June 2025
- Published by Frontiers Media SA in Frontiers in Sports and Active Living
- Vol. 7, 1627820
Performance tracking in female youth soccer through wearables and subjective assessments
This study investigates performance development and the relationship between subjective and objective training assessments in female youth soccer using wearable sensor technology. The aim of this study was to assess how subjective post-training ratings (intensity and happiness) relate to high-percentile performance outputs, and to identify longitudinal trends in female youth soccer players using IMU-based wearable data. Data were collected over a 14-month period from 46 players (U17 and U20 teams) equipped with foot-mounted inertial measurement units (IMUs) during regular training sessions. Objective performance metrics, including 95th percentile of ball speed, peak speed, and absolute distance, were derived using a multi-stage machine learning pipeline, while subjective metrics (intensity and happiness) were collected via post-session Likert-scale questionnaires using an app. Using the modified Mann-Kendall test, we found 30 significant longitudinal trends, with 14 positive and 16 negative trends across key performance metrics. Peak speed showed the highest number of trends (13), followed by absolute distance (10) and ball speed (7). Correlation analyses based on the Spearman coefficient (with False Discovery Rate correction) revealed meaningful associations between subjective self-assessments and high-percentile performance metrics, with notable differences across player positions and age groups. A robustness check confirmed these patterns also hold when analyzing the 99th percentile of performance outputs. Our findings underscore the value of combining wearable sensor data with subjective evaluations for individualized, role-specific performance monitoring and training optimization in youth soccer. However, as an exploratory study with a single cohort, findings require further validation in broader populations.
- 30 June 2025
- Published by Frontiers Media SA in Frontiers in Sports and Active Living
- Vol. 7, 1627820
Open AccessPerformance tracking in female youth soccer through wearables and subjective assessments
This study investigates performance development and the relationship between subjective and objective training assessments in female youth soccer using wearable sensor technology. The aim of this study was to assess how subjective post-training ratings (intensity and happiness) relate to high-percentile performance outputs, and to identify longitudinal trends in female youth soccer players using IMU-based wearable data. Data were collected over a 14-month period from 46 players (U17 and U20 teams) equipped with foot-mounted inertial measurement units (IMUs) during regular training sessions. Objective performance metrics, including 95th percentile of ball speed, peak speed, and absolute distance, were derived using a multi-stage machine learning pipeline, while subjective metrics (intensity and happiness) were collected via post-session Likert-scale questionnaires using an app. Using the modified Mann-Kendall test, we found 30 significant longitudinal trends, with 14 positive and 16 negative trends across key performance metrics. Peak speed showed the highest number of trends (13), followed by absolute distance (10) and ball speed (7). Correlation analyses based on the Spearman coefficient (with False Discovery Rate correction) revealed meaningful associations between subjective self-assessments and high-percentile performance metrics, with notable differences across player positions and age groups. A robustness check confirmed these patterns also hold when analyzing the 99th percentile of performance outputs. Our findings underscore the value of combining wearable sensor data with subjective evaluations for individualized, role-specific performance monitoring and training optimization in youth soccer. However, as an exploratory study with a single cohort, findings require further validation in broader populations.
- Open Access
- 18 July 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (7) , e0307255
Development of a simple algorithm to detect big air jumps and jumps during skiing
Jumping is an important task in skiing, snowboarding, ski jumping, figure skating, volleyball and many other sports. In these examples, jumping tasks are a performance criterion, and therefore detailed insight into them is important for athletes and coaches. Therefore, this paper aims to introduce a simple and easy-to-implement jump detection algorithm for skiing using acceleration data from inertial measurement units attached to ski boots. The algorithm uses the average of the absolute vertical accelerations of the two boots. We provide results for different parameter settings of the algorithm and two types of jumps: Big Air jumps and jumps during skiing. The latter are divided into small (time of flight < 500 ms) and medium (time of flight ≥ 500 ms) jumps. The algorithm detects 100% of Big Air, 94% of medium and 44% of small jumps. In addition, the settings with the highest detection rates also have the highest number of overdetected jumps. To resolve this conflict, a penalty-adjusted score that considers the number of overdetected jumps in the final performance analysis is proposed.
- 18 July 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (7) , e0307255
Open AccessDevelopment of a simple algorithm to detect big air jumps and jumps during skiing
Jumping is an important task in skiing, snowboarding, ski jumping, figure skating, volleyball and many other sports. In these examples, jumping tasks are a performance criterion, and therefore detailed insight into them is important for athletes and coaches. Therefore, this paper aims to introduce a simple and easy-to-implement jump detection algorithm for skiing using acceleration data from inertial measurement units attached to ski boots. The algorithm uses the average of the absolute vertical accelerations of the two boots. We provide results for different parameter settings of the algorithm and two types of jumps: Big Air jumps and jumps during skiing. The latter are divided into small (time of flight < 500 ms) and medium (time of flight ≥ 500 ms) jumps. The algorithm detects 100% of Big Air, 94% of medium and 44% of small jumps. In addition, the settings with the highest detection rates also have the highest number of overdetected jumps. To resolve this conflict, a penalty-adjusted score that considers the number of overdetected jumps in the final performance analysis is proposed.
- Open Access
- 1 September 2023
- Published by Journal of Sports Science and Medicine in Journal of Sports Science and Medicine
- Vol. 22 (3) , 476-487
Predicting Injury and Illness with Machine Learning in Elite Youth Soccer: A Comprehensive Monitoring Approach over 3 Months
The search for monitoring tools that provide early indication of injury and illness could contribute to better player protection. The aim of the present study was to i) determine the feasibility of and adherence to our monitoring approach, and ii) identify variables associated with up-coming illness and injury. We incorporated a comprehensive set of monitoring tools consisting of external load and physical fitness data, questionnaires, blood, neuromuscular-, hamstring, hip abductor and hip adductor performance tests performed over a three-month period in elite under-18 academy soccer players. Twenty-five players (age: 16.6 ± 0.9 years, height: 178 ± 7 cm, weight: 74 ± 7 kg, VO2max: 59 ± 4 ml/min/kg) took part in the study. In addition to evaluating adherence to the monitoring approach, data were analyzed using a linear support vector machine (SVM) to predict illness and injuries. The approach was feasible, with no injuries or dropouts due to the monitoring process. Questionnaire adherence was high at the beginning and decreased steadily towards the end of the study. An SVM resulted in the best classification results for three classification tasks, i.e., illness prediction, illness determination and injury prediction. For injury prediction, one of four injuries present in the test data set was detected, with 96.3% of all data points (i.e., injuries and non-injuries) correctly detected. For both illness prediction and determination, there was only one illness in the test data set that was detected by the linear SVM. However, the model showed low precision for injury and illness prediction with a considerable number of false-positives. The results demonstrate the feasibility of a holistic monitoring approach with the possibility of predicting illness and injury. Additional data points are needed to improve the prediction models. In practical application, this may lead to overcautious recommendations on when players should be protected from injury and illness.
- 1 September 2023
- Published by Journal of Sports Science and Medicine in Journal of Sports Science and Medicine
- Vol. 22 (3) , 476-487
Open AccessPredicting Injury and Illness with Machine Learning in Elite Youth Soccer: A Comprehensive Monitoring Approach over 3 Months
The search for monitoring tools that provide early indication of injury and illness could contribute to better player protection. The aim of the present study was to i) determine the feasibility of and adherence to our monitoring approach, and ii) identify variables associated with up-coming illness and injury. We incorporated a comprehensive set of monitoring tools consisting of external load and physical fitness data, questionnaires, blood, neuromuscular-, hamstring, hip abductor and hip adductor performance tests performed over a three-month period in elite under-18 academy soccer players. Twenty-five players (age: 16.6 ± 0.9 years, height: 178 ± 7 cm, weight: 74 ± 7 kg, VO2max: 59 ± 4 ml/min/kg) took part in the study. In addition to evaluating adherence to the monitoring approach, data were analyzed using a linear support vector machine (SVM) to predict illness and injuries. The approach was feasible, with no injuries or dropouts due to the monitoring process. Questionnaire adherence was high at the beginning and decreased steadily towards the end of the study. An SVM resulted in the best classification results for three classification tasks, i.e., illness prediction, illness determination and injury prediction. For injury prediction, one of four injuries present in the test data set was detected, with 96.3% of all data points (i.e., injuries and non-injuries) correctly detected. For both illness prediction and determination, there was only one illness in the test data set that was detected by the linear SVM. However, the model showed low precision for injury and illness prediction with a considerable number of false-positives. The results demonstrate the feasibility of a holistic monitoring approach with the possibility of predicting illness and injury. Additional data points are needed to improve the prediction models. In practical application, this may lead to overcautious recommendations on when players should be protected from injury and illness.
- Open Access
- 1 September 2023
- Published by Journal of Sports Science and Medicine in Journal of Sports Science and Medicine
- Vol. 22 (3) , 475-486
Predicting Injury and Illness with Machine Learning in Elite Youth Soccer: A Comprehensive Monitoring Approach over 3 Months
The search for monitoring tools that provide early indication of injury and illness could contribute to better player protection. The aim of the present study was to i) determine the feasibility of and adherence to our monitoring approach, and ii) identify variables associated with up-coming illness and injury. We incorporated a comprehensive set of monitoring tools consisting of external load and physical fitness data, questionnaires, blood, neuromuscular-, hamstring, hip abductor and hip adductor performance tests performed over a three-month period in elite under-18 academy soccer players. Twenty-five players (age: 16.6 ± 0.9 years, height: 178 ± 7 cm, weight: 74 ± 7 kg, VO2max: 59 ± 4 ml/min/kg) took part in the study. In addition to evaluating adherence to the monitoring approach, data were analyzed using a linear support vector machine (SVM) to predict illness and injuries. The approach was feasible, with no injuries or dropouts due to the monitoring process. Questionnaire adherence was high at the beginning and decreased steadily towards the end of the study. An SVM resulted in the best classification results for three classification tasks, i.e., illness prediction, illness determination and injury prediction. For injury prediction, one of four injuries present in the test data set was detected, with 96.3% of all data points (i.e., injuries and non-injuries) correctly detected. For both illness prediction and determination, there was only one illness in the test data set that was detected by the linear SVM. However, the model showed low precision for injury and illness prediction with a considerable number of false-positives. The results demonstrate the feasibility of a holistic monitoring approach with the possibility of predicting illness and injury. Additional data points are needed to improve the prediction models. In practical application, this may lead to overcautious recommendations on when players should be protected from injury and illness.
- 1 September 2023
- Published by Journal of Sports Science and Medicine in Journal of Sports Science and Medicine
- Vol. 22 (3) , 475-486
Open AccessPredicting Injury and Illness with Machine Learning in Elite Youth Soccer: A Comprehensive Monitoring Approach over 3 Months
The search for monitoring tools that provide early indication of injury and illness could contribute to better player protection. The aim of the present study was to i) determine the feasibility of and adherence to our monitoring approach, and ii) identify variables associated with up-coming illness and injury. We incorporated a comprehensive set of monitoring tools consisting of external load and physical fitness data, questionnaires, blood, neuromuscular-, hamstring, hip abductor and hip adductor performance tests performed over a three-month period in elite under-18 academy soccer players. Twenty-five players (age: 16.6 ± 0.9 years, height: 178 ± 7 cm, weight: 74 ± 7 kg, VO2max: 59 ± 4 ml/min/kg) took part in the study. In addition to evaluating adherence to the monitoring approach, data were analyzed using a linear support vector machine (SVM) to predict illness and injuries. The approach was feasible, with no injuries or dropouts due to the monitoring process. Questionnaire adherence was high at the beginning and decreased steadily towards the end of the study. An SVM resulted in the best classification results for three classification tasks, i.e., illness prediction, illness determination and injury prediction. For injury prediction, one of four injuries present in the test data set was detected, with 96.3% of all data points (i.e., injuries and non-injuries) correctly detected. For both illness prediction and determination, there was only one illness in the test data set that was detected by the linear SVM. However, the model showed low precision for injury and illness prediction with a considerable number of false-positives. The results demonstrate the feasibility of a holistic monitoring approach with the possibility of predicting illness and injury. Additional data points are needed to improve the prediction models. In practical application, this may lead to overcautious recommendations on when players should be protected from injury and illness.
- Open Access
- 27 July 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (15) , 8684
Classification of Human Motion Data Based on Inertial Measurement Units in Sports: A Scoping Review
Inertial measurement units (IMU) are widely used in sports applications to digitise human motion by measuring acceleration and rotational velocity in three-dimensional space. A common machine learning problem is the classification of human motion primitives from IMU data. In order to investigate the classification methods used in the existing literature and to analyse whether and how the time-dependent data structure is considered in the classification process of motion data analysis in sports, a scoping review was conducted. Based on a keyword search, articles from 2010 to 2021 were extracted, and 93 articles were relevant for data extraction. Over- and undersampling of data and data augmentation techniques were rarely used. The classification methods applied can be divided into three main branches: classic machine learning and deep learning models, threshold-based approaches, and dynamic time warping. The most often applied algorithms were support vector machines (SVM), followed by neural networks and k-nearest neighbours. In comparative works, when more than one classifier was applied, random forests, neural networks, boosting models and SVM were found to be the methods that achieved the highest accuracy. If the time-dependent data structure was taken into account, it was incorporated either within the models, for example, by using long-short-term memory models or within the feature calculation step by using rolling windows with an overlap, which was the most common method of considering the time dependency of the IMU data.
- 27 July 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (15) , 8684
Open AccessClassification of Human Motion Data Based on Inertial Measurement Units in Sports: A Scoping Review
Inertial measurement units (IMU) are widely used in sports applications to digitise human motion by measuring acceleration and rotational velocity in three-dimensional space. A common machine learning problem is the classification of human motion primitives from IMU data. In order to investigate the classification methods used in the existing literature and to analyse whether and how the time-dependent data structure is considered in the classification process of motion data analysis in sports, a scoping review was conducted. Based on a keyword search, articles from 2010 to 2021 were extracted, and 93 articles were relevant for data extraction. Over- and undersampling of data and data augmentation techniques were rarely used. The classification methods applied can be divided into three main branches: classic machine learning and deep learning models, threshold-based approaches, and dynamic time warping. The most often applied algorithms were support vector machines (SVM), followed by neural networks and k-nearest neighbours. In comparative works, when more than one classifier was applied, random forests, neural networks, boosting models and SVM were found to be the methods that achieved the highest accuracy. If the time-dependent data structure was taken into account, it was incorporated either within the models, for example, by using long-short-term memory models or within the feature calculation step by using rolling windows with an overlap, which was the most common method of considering the time dependency of the IMU data.
- Open Access
- 18 May 2023
- Published by Springer Nature in Behaviormetrika
- Vol. 51 (1) , 341-358
Generalisability of sleep stage classification based on interbeat intervals: validating three machine learning approaches on self-recorded test data
Classifying sleep stages is an important basis for neuroscience, health sciences, psychology and many other fields. However, the manual determination of sleep stages is tedious and time consuming. Therefore, the development of automatic sleep stage classifiers based on data collected with low-cost sensor systems is an important research area. This study aims to analyse the generalisability of different machine learning approaches for sleep stage classification. We train three different models (random forest, CNN-LSTM and seq2seq) for classifying three as well as four sleep stages, with the MESA data set. For validation, we use a fivefold cross-validation and further validate the models with one new self-recorded test data set to analyse the models’ generalisability to a completely new cohort with different characteristics with regard to age and health status. Our results show that the two deep learning approaches performed better than the random forest. Moreover, all models are generalisable and therefore suitable for sleep stage classification on a new three-stage classification data set. However, generalisability for the four-stage classification task shows poorer performance, and therefore requires new approaches such as transfer learning or a larger data set to train the models.
- 18 May 2023
- Published by Springer Nature in Behaviormetrika
- Vol. 51 (1) , 341-358
Open AccessGeneralisability of sleep stage classification based on interbeat intervals: validating three machine learning approaches on self-recorded test data
Classifying sleep stages is an important basis for neuroscience, health sciences, psychology and many other fields. However, the manual determination of sleep stages is tedious and time consuming. Therefore, the development of automatic sleep stage classifiers based on data collected with low-cost sensor systems is an important research area. This study aims to analyse the generalisability of different machine learning approaches for sleep stage classification. We train three different models (random forest, CNN-LSTM and seq2seq) for classifying three as well as four sleep stages, with the MESA data set. For validation, we use a fivefold cross-validation and further validate the models with one new self-recorded test data set to analyse the models’ generalisability to a completely new cohort with different characteristics with regard to age and health status. Our results show that the two deep learning approaches performed better than the random forest. Moreover, all models are generalisable and therefore suitable for sleep stage classification on a new three-stage classification data set. However, generalisability for the four-stage classification task shows poorer performance, and therefore requires new approaches such as transfer learning or a larger data set to train the models.
- Open Access
- 22 March 2023
- Published by Springer Nature in Sports Engineering
- Vol. 26 (1) , 1-10
Wearable chest sensor for stride and respiration detection during running
Endurance running is among the most popular physical activities partially due to its low barriers to entry. However, some people avoid running because of respiratory distress, and respiratory monitoring could help prevent this. Wearable sensors are valuable for respiration detection during exercise and enable respiratory feedback in real time. Therefore, this study presents a wearable chest-mounted stride and respiration sensor including step and flow reversal event detection algorithms. The algorithms were evaluated using precision and recall between detected and reference events with respect to different levels of breathing depth, motion artifact, thoracic skin temperature and sweat. Overall F1 scores reached 93.2%, 97.4% and 97.2% for step, expiration and inspiration events, respectively. No significant effect on event detection performance was observed for breathing depth, stride motion artifact, or thoracic skin temperature. In contrast, sweat level slightly decreased detection performance. Consequently, this sensor is able to accurately measure stride and respiration during running and could be suitable for use as a system to guide runners’ respiration during exercise.
- 22 March 2023
- Published by Springer Nature in Sports Engineering
- Vol. 26 (1) , 1-10
Open AccessWearable chest sensor for stride and respiration detection during running
Endurance running is among the most popular physical activities partially due to its low barriers to entry. However, some people avoid running because of respiratory distress, and respiratory monitoring could help prevent this. Wearable sensors are valuable for respiration detection during exercise and enable respiratory feedback in real time. Therefore, this study presents a wearable chest-mounted stride and respiration sensor including step and flow reversal event detection algorithms. The algorithms were evaluated using precision and recall between detected and reference events with respect to different levels of breathing depth, motion artifact, thoracic skin temperature and sweat. Overall F1 scores reached 93.2%, 97.4% and 97.2% for step, expiration and inspiration events, respectively. No significant effect on event detection performance was observed for breathing depth, stride motion artifact, or thoracic skin temperature. In contrast, sweat level slightly decreased detection performance. Consequently, this sensor is able to accurately measure stride and respiration during running and could be suitable for use as a system to guide runners’ respiration during exercise.