Real-time forecasting of exercise-induced fatigue from wearable sensors

Jiang, Yanran; Malliaras, Peter; Chen, Bernard; Kulić, Dana

doi:10.1016/j.compbiomed.2022.105905

Cited by 13 publications

(5 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [ 21 ], a neural network mixed with Fourier transform was used, achieving an RMSE value of 0.134, predicting a total of 0.6 s ahead using 1 s of input data. Meanwhile, in [ 22 ], an adversarial training based on a transformer Generator, a CNN Critic Network, and an action classifier achieved an RMSE value of 0.180 between the forecast and ground truth signals in fatigue prediction future 0.8 s. Compared to these works, our model can handle multivariate signal forecasting signals up to 2.56 s compared to prior studies. Due to the majority of prior works for time series, forecasters are used in applications such as weather, asset prediction, or data augmentation, and the lack of forecasters used for human activity prediction is not possible to directly compare prior models with the structure proposed in this work.…”

Section: Discussionmentioning

confidence: 99%

“…Although this model could obtain intention prediction results, it is related to the complete sequence of actions followed by the subject and not only the current activity. In [ 22 ], a multiclass fatigue prediction up to 0.8 s is carried out using the forecasted IMU signals, achieving an average accuracy of 83% with a correlation coefficient of 92%. Compared to these prior works, our HAP system only uses 2.56 s of IMU data instead of the hold record to perform the activity prediction, achieving an average accuracy of 97.981% and a precision improvement of 3.13%.…”

Section: Discussionmentioning

confidence: 99%

“…Although sensor-based HAR systems have been popular due to the growth in wearable devices in daily life [ 18 , 19 ], sensor-based HAP works are rare and have not been actively investigated. Recently, multivariate motion signal predictions from IMU sensors have been proposed through attention models, capable of retaining pattern and feature information across multiple channels [ 20 , 21 , 22 , 23 , 24 ]. Therefore, these forecasting models on time series data can be used in HAP to generate future activity signals, making HAP feasible by classifying the forecast signals.…”

Section: Introductionmentioning

confidence: 99%

“…The system was evaluated in a fall detection study, predicting a fall up to 0.6 s before the event. In [ 22 ], an adversarial network based on a transformer model and attention layers was used to forecast motion signals and predict the fatigue level during sports training routines. They obtained a Pearson correlation of 0.92 and a fatigue accuracy of 83%.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Human Activity Prediction Based on Forecasted IMU Activity Signals by Sequence-to-Sequence Deep Neural Networks

Jaramillo

Chola

Jeong

et al. 2023

Sensors

View full text Add to dashboard Cite

Human Activity Recognition (HAR) has gained significant attention due to its broad range of applications, such as healthcare, industrial work safety, activity assistance, and driver monitoring. Most prior HAR systems are based on recorded sensor data (i.e., past information) recognizing human activities. In fact, HAR works based on future sensor data to predict human activities are rare. Human Activity Prediction (HAP) can benefit in multiple applications, such as fall detection or exercise routines, to prevent injuries. This work presents a novel HAP system based on forecasted activity data of Inertial Measurement Units (IMU). Our HAP system consists of a deep learning forecaster of IMU activity signals and a deep learning classifier to recognize future activities. Our deep learning forecaster model is based on a Sequence-to-Sequence structure with attention and positional encoding layers. Then, a pre-trained deep learning Bi-LSTM classifier is used to classify future activities based on the forecasted IMU data. We have tested our HAP system for five daily activities with two tri-axial IMU sensors. The forecasted signals show an average correlation of 91.6% to the actual measured signals of the five activities. The proposed HAP system achieves an average accuracy of 97.96% in predicting future activities.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Human Activity Prediction Based on Forecasted IMU Activity Signals by Sequence-to-Sequence Deep Neural Networks

Jaramillo

Chola

Jeong

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…These intensity levels can be monitored using a set of wearables proposed in this work, allowing exercise intensity to be precisely modulated during treatment sessions 22 . Finally, suitable AI-based approach can provide real-time and personalized fatigue management during the rehabilitation process 92 .…”

Section: Usage Notesmentioning

confidence: 99%

A dataset for fatigue estimation during shoulder internal and external rotation movements using wearables

Yasar,

Sica,

O’Flynn

et al. 2024

Sci Data

View full text Add to dashboard Cite

Wearable sensors have recently been extensively used in sports science, physical rehabilitation, and industry providing feedback on physical fatigue. Information obtained from wearable sensors can be analyzed by predictive analytics methods, such as machine learning algorithms, to determine fatigue during shoulder joint movements, which have complex biomechanics. The presented dataset aims to provide data collected via wearable sensors during a fatigue protocol involving dynamic shoulder internal rotation (IR) and external rotation (ER) movements. Thirty-four healthy subjects performed shoulder IR and ER movements with different percentages of maximal voluntary isometric contraction (MVIC) force until they reached the maximal exertion. The dataset includes demographic information, anthropometric measurements, MVIC force measurements, and digital data captured via surface electromyography, inertial measurement unit, and photoplethysmography, as well as self-reported assessments using the Borg rating scale of perceived exertion and the Karolinska sleepiness scale. This comprehensive dataset provides valuable insights into physical fatigue assessment, allowing the development of fatigue detection/prediction algorithms and the study of human biomechanical characteristics during shoulder movements within a fatigue protocol.

show abstract

Intelligent fatigue detection based on hierarchical multi-scale ECG representations and HRV measures

Mu,

Liao,

Tao

et al. 2024

Biomedical Signal Processing and Control

View full text Add to dashboard Cite

Real-time forecasting of exercise-induced fatigue from wearable sensors

Cited by 13 publications

References 24 publications

Human Activity Prediction Based on Forecasted IMU Activity Signals by Sequence-to-Sequence Deep Neural Networks

Human Activity Prediction Based on Forecasted IMU Activity Signals by Sequence-to-Sequence Deep Neural Networks

A dataset for fatigue estimation during shoulder internal and external rotation movements using wearables

Intelligent fatigue detection based on hierarchical multi-scale ECG representations and HRV measures

Contact Info

Product

Resources

About