On the Impact of Biceps Muscle Fatigue in Human Activity Recognition

Abstract

Nowadays, Human Activity Recognition (HAR) systems, which use wearables and smart systems, are a part of our daily life. Despite the abundance of literature in the area, little is known about the impact of muscle fatigue on these systems’ performance. In this work, we use the biceps concentration curls exercise as an example of a HAR activity to observe the impact of fatigue impact on such systems. Our dataset consists of 3000 biceps concentration curls performed and collected from 20 volunteers aged between 20–35. Our findings indicate that fatigue often occurs in later sets of an exercise and extends the completion time of later sets by up to 31% and decreases muscular endurance by 4.1%. Another finding shows that changes in data patterns are often occurring during fatigue presence, causing seven features to become statistically insignificant. Further findings indicate that fatigue can cause a substantial decrease in performance in both subject-specific and cross-subject models. Finally, we observed that a Feedforward Neural Network (FNN) showed the best performance in both cross-subject and subject-specific models in all our evaluations.

Toward the Personalization of Biceps Fatigue Detection Model for Gym Activity: An Approach to Utilize Wearables’ Data from the Crowd

“…The second reason is that medium-weight dumbbells are often reported as the most commonly used dumbbells across gym-goers [54]. Third, in previous work, we found that the 4.5 kg weight dumbbell provides the best trade-off between number data points recorded in data sessions and time to reach fatigue during an exercise [55]. We use Borg's scale in our work because we believe that RPE is an appropriate marker of fatigue as previous studies within sport science have proven that RPE is capable of modeling a person's performance better in the real world compared to only heart rate monitoring [7,49,56].…”

“…A recent study shows a significant performance improvement (from 89.83% to 96.62%) after utilizing deep learning in HAR systems [107]. We opt to use DT and ANN models in the current study to examine whether our personalization approach can mitigate the hindering effect of subject data variability encountered in our previous works [45,55]. So, since our findings seem encouraging, we have decided to use CNN in our future work to examine our personalization approach further.…”

“…Besides the eight features mentioned before, we also select three other features often associated with fatigue for better performance: skewness, IoP, and MSP [76][77][78][79]. A previous study suggests considering the skewness of the data when detecting fatigue in repetitive muscle movements such as bicep curls [55]. We select skewness as a fatigue feature because, during the repetitions' extraction and labeling process, we observed the following: (1) Non-fatigue repetitions are relatively symmetrical during the repetitions' extraction and labeling process.…”

“…The current work aims to mitigate the hindering effect of subject data variability from our prior works based on the Borg scale [45,55]. To achieve fairness in evaluation and continuity of our prior works, we use the Borg scale in our personalization approach.…”

Data-Driven Approach for Upper Limb Fatigue Estimation Based on Wearable Sensors

“…Leave-one-out cross-validation (LOOCV) was used, where the number of folds equals the number of instances or subjects in the training set. In this case, 70% is equivalent to 21 subjects in the training data [74,75].…”