Multimodal Human Activity Recognition From Wearable Inertial Sensors Using Machine Learning

Badawi, Abeer; Al-Kabbany, Ahmad; Shaban, Heba

doi:10.1109/iecbes.2018.8626737

Cited by 18 publications

(14 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our own previous research [5] investigates the relation between the recognition accuracy and the IMU's placementsand-orientations [5]. We also highlight the impact (in terms of accuracy) of applying feature selection techniques on the IMU's data [6].…”

Section: Introductionmentioning

confidence: 93%

“…is dataset contains twelve activities, with the following ID numbers: walking with ID number [0], running with ID number [1], going down with ID number [2], going up with ID number [3], sitting with ID number [4], sitting down with ID number [5], standing up with ID number [6], standing with ID number [7], bicycling with ID number [8], down by elevator with ID number [9], up by elevator with ID number [10], and sitting in the car with ID number [11]. e data was collected from 18 adults (14 males and 4 females).…”

Section: Dataset Descriptionmentioning

confidence: 99%

“…To evaluate the best sensor placement, we considered the accelerometer and gyroscope sensors only, because of their ability to recognize the activities accurately as mentioned in our previous work [5]. e classification techniques were applied on the dataset of the inertial sensor placements, axes, and magnitude, in addition to the EMG sensor.…”

Section: Performance Evaluation For Optimal Sensor Placementmentioning

confidence: 99%

“…Particularly, we incorporate our knowledge on the best sensor placements-andorientations, in addition to our findings on the impact of feature selection, and we show that EMG data can influence the system's recognition capacity positively. In this work, we have considered widely used classifiers, namely, support vector machine (SVM), naive Bayes (NB), neural networks as multilayer perceptron (MLP), decision tree (DT), k-nearest neighbor (k-NN), and random forest (RF) [5]. We have also followed the study in [6] while applying feature selection.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Sensor Type, Axis, and Position-Based Fusion and Feature Selection for Multimodal Human Daily Activity Recognition in Wearable Body Sensor Networks

Badawi

Al-Kabbany

Shaban

2020

Journal of Healthcare Engineering

Self Cite

View full text Add to dashboard Cite

This research addresses the challenge of recognizing human daily activities using surface electromyography (sEMG) and wearable inertial sensors. Effective and efficient recognition in this context has emerged as a cornerstone in robust remote health monitoring systems, among other applications. We propose a novel pipeline that can attain state-of-the-art recognition accuracies on a recent-and-standard dataset—the Human Gait Database (HuGaDB). Using wearable gyroscopes, accelerometers, and electromyography sensors placed on the thigh, shin, and foot, we developed an approach that jointly performs sensor fusion and feature selection. Being done jointly, the proposed pipeline empowers the learned model to benefit from the interaction of features that might have been dropped otherwise. Using statistical and time-based features from heterogeneous signals of the aforementioned sensor types, our approach attains a mean accuracy of 99.8%, which is the highest accuracy on HuGaDB in the literature. This research underlines the potential of incorporating EMG signals especially when fusion and selection are done simultaneously. Meanwhile, it is valid even with simple off-the-shelf feature selection methods such the Sequential Feature Selection family of algorithms. Moreover, through extensive simulations, we show that the left thigh is a key placement for attaining high accuracies. With one inertial sensor on that single placement alone, we were able to achieve a mean accuracy of 98.4%. The presented in-depth comparative analysis shows the influence that every sensor type, position, and placement can have on the attained recognition accuracies—a tool that can facilitate the development of robust systems, customized to specific scenarios and real-life applications.

show abstract

Section: Introductionmentioning

confidence: 93%

Section: Dataset Descriptionmentioning

confidence: 99%

Section: Performance Evaluation For Optimal Sensor Placementmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Sensor Type, Axis, and Position-Based Fusion and Feature Selection for Multimodal Human Daily Activity Recognition in Wearable Body Sensor Networks

Badawi

Al-Kabbany

Shaban

2020

Journal of Healthcare Engineering

Self Cite

View full text Add to dashboard Cite

show abstract

“…More recently, the deployment of machine learning (ML) techniques in addition to wearable sensors has been evolving [19]. The use of ML algorithms enables for both accurate and automatic assessment of patient symptoms, which in turn deemed to help in both decision making and long-term monitoring of patients.…”

Section: Ralated Workmentioning

confidence: 99%

Wearable Accelerometer and sEMG-Based Upper Limb BSN for Tele-Rehabilitation

et al. 2019

Self Cite

View full text Add to dashboard Cite

Assessment of human locomotion using wearable sensors is an efficient way of getting useful information about human health status, and determining human locomotion abnormalities. Wearable sensors do not only provide the opportunity to assess the behavior of patients as it happens in their daily life activities, but also provide quantitative, meaningful feedback data of patients to their therapists. This can pinpoint the cause of problems and help in maximizing their recovery rates. The popularity of using wearable sensors has received attention from a number of researchers from both the academic and industrial fields in the past few years. The different types of wearable sensors have given birth to the realization of a standard measurement model that can support different types of applications. Wireless body area networks (WBANs) are starting to replace traditional healthcare systems by enabling long-term monitoring of patients and tele-rehabilitation, especially those who suffer from chronic diseases. This paper investigates using wearable accelerometers and surface electromyography (EMG) in human locomotion monitoring for tele-rehabilitation. It proposes and investigates new positions for the proposed sensors, and compares the measured signals to similar techniques proposed in the literature. Realistic measurements show that the proposed positions of surface EMG sensors (on the forearm muscles) provide more reliable results in the classification of motion abnormality as compared to the sensor positions proposed in the literature (biceps muscles). Seven statistical features were extracted from accelerometer signals, and four time domain (TD) features are extracted from EMG signals. These features are used to construct six machine learning classifiers for automatic classification of Parkinson’s tremor. These models include; decision tree (DT), linear discriminant analysis analysis (LDA), k-nearest-neighbor (kNN), support vector machine (SVM), boosted tree and bagged tree classifiers. The performance of the applied classifiers is analyzed using accuracy, confusion matrix, and area under ROC (AUC) curve. The results are also compared to corresponding findings in the literature. The experimental results show that the highest classification accuracy is achieved when using the proposed measurement set and bagged tree classifier with a value of 99.6%.

show abstract

Advances in Biosignal Sensing and Signal Processing Methods with Wearable Devices

2022

View full text Add to dashboard Cite

Analysis & Sensing www.analysis-sensing.org Review doi.org/10.1002/anse.202200062Wearable devices have received significant attention recently for their ability to monitor critical physiological signals noninvasively, such as electrocardiography, electroencephalography, electromyography, and photoplethysmography. These biointegrated wearable systems can potentially fill gaps in conventional clinical practice by providing highly cost-effective health characterization and portable continuous health monitoring. Further, the physiological signals measured by wearables require post-processing to derive meaningful values, such as heart rate or blood oxygen saturation. This requirement, in conjunction with the smaller form factor and limited sensor count of the miniaturized systems, often necessitates robust signal processing and data analysis to approach the stringent performance specifications of conventional medical devices, and machine learning techniques have found success in filling this analytical role for their ability to learn complex functional relationships. Thus, this review outlines a systematic summary of the latest research on various wearable devices and their biosignal sensing and signal processing methods, emphasizing machine learning. We also discuss the developmental challenges and advantages of current machine-learning methods, while suggesting research directions for future studies.

show abstract

Multimodal Human Activity Recognition From Wearable Inertial Sensors Using Machine Learning

Cited by 18 publications

References 10 publications

Sensor Type, Axis, and Position-Based Fusion and Feature Selection for Multimodal Human Daily Activity Recognition in Wearable Body Sensor Networks

Sensor Type, Axis, and Position-Based Fusion and Feature Selection for Multimodal Human Daily Activity Recognition in Wearable Body Sensor Networks

Wearable Accelerometer and sEMG-Based Upper Limb BSN for Tele-Rehabilitation

Advances in Biosignal Sensing and Signal Processing Methods with Wearable Devices

Contact Info

Product

Resources

About