Article
Chemistry, Analytical
Thomas Perri, Machar Reid, Alistair Murphy, Kieran Howle, Rob Duffield
Summary: This study evaluated the accuracy of tennis-specific stroke and movement event detection algorithms using a wearable sensor. The results showed high accuracy in detecting strokes, particularly serves, but difficulties in detecting volleys. The most common tennis-specific footwork patterns were identified as active movements and linear running.
Article
Computer Science, Information Systems
Anastasia Motrenko, Egor Simchuk, Renat Khairullin, Andrey Inyakin, Daniil Kashirin, Vadim Strijov
Summary: The paper addresses the problem of human activity recognition based on data from wearable sensors, proposing a hierarchical representation of activities as sets of low-level actions. This approach allows for a more condensed exploration of activities, overcoming the limitations of time series analysis.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Kevin Chapron, Patrick Lapointe, Isabelle Lessard, Hans Darsmstadt-Belanger, Kevin Bouchard, Cynthia Gagnon, Melissa Lavoie, Elise Duchesne, Sebastien Gaboury
Summary: The study developed an assistive system that can identify activities related to physical exercise and help patients with training, achieving encouraging results.
Article
Engineering, Industrial
A. Mastakouris, G. Andriosopoulou, D. Masouros, P. Benardos, G. -c. Vosniakos, D. Soudris
Summary: This work presents a deep learning based approach for automated activity recognition of two human workers in a production floor environment. By using a smartphone as a wearable sensor and employing neural networks for classification, the activities of the workers can be accurately identified in real time.
JOURNAL OF MANUFACTURING SYSTEMS
(2023)
Article
Medicine, General & Internal
Frederick K. Ho, Fanny Petermann-Rocha, Solange Parra-Soto, Jirapitcha Boonpor, Jason M. R. Gill, Stuart R. Gray, Jill P. Pell, Carlos Celis-Morales
Summary: This study using prospective cohort data found an association betweendevice-measured moderate and vigorous physical activity (MPA and VPA) and lower risk of affective disorders. Assuming causality, achieving 150 minutes per week of MPA and 75 minutes per week of VPA could prevent 5.14% and 18.88% of affective disorders, respectively.
Article
Computer Science, Artificial Intelligence
Duygu Bagci Das, Derya Birant
Summary: This study proposes a novel method, HAR-MIL, for human activity recognition based on multi-instance learning. The method represents human activities as bags of various wearable sensors and provides a flexible model by eliminating the restrictions of traditional single-instance representation. Experimental results show that HAR-MIL is effective for wearable sensor-based HAR with high classification accuracy.
Article
Computer Science, Information Systems
Chunjing Xiao, Shiming Chen, Fan Zhou, Jie Wu
Summary: This article introduces a self-supervised few-shot time-series segmentation framework (SFTSeg) for extracting valuable activity segments from sensor data. By incorporating few-shot learning and line-level data augmentation, the framework enables activity segmentation with only a few labeled target samples and achieves significant improvements in segmentation performance.
IEEE TRANSACTIONS ON MOBILE COMPUTING
(2023)
Article
Engineering, Electrical & Electronic
Mahdi Pedram, Ramesh Kumar Sah, Seyed Ali Rokni, Marjan Nourollahi, Hassan Ghasemzadeh
Summary: Advances in embedded systems have led to the integration of wearable sensors in health monitoring. However, due to the personalized nature of human movement and the limitations of embedded sensors, a resource-efficient framework is needed for real-time activity recognition.
IEEE SENSORS JOURNAL
(2022)
Article
Public, Environmental & Occupational Health
Tristan Stampfler, Mohamed Elgendi, Richard Ribon Fletcher, Carlo Menon
Summary: The field of digital phenotyping uses smartphone sensors to understand users' psychological state and behavior, improving health support systems. This article presents a deep learning method using the Resnet architecture to recognize human behavior, achieving higher accuracy and F1-score than existing techniques. The authors also discuss future research directions and real-time implementation of their approach for behavior recognition.
FRONTIERS IN PUBLIC HEALTH
(2023)
Article
Biophysics
John J. Davis, Marcin Straczkiewicz, Jaroslaw Harezlak, Allison H. Gruber
Summary: The study developed an algorithm to identify running bouts in accelerometer data with high accuracy, ranging from 98.4% to 99.4%. The CARL classifier can accurately detect running bouts as short as three seconds in free-living accelerometry data.
PHYSIOLOGICAL MEASUREMENT
(2021)
Article
Chemistry, Analytical
Mohamed Elshafei, Diego Elias Costa, Emad Shihab
Summary: This research investigates the impact of muscle fatigue on Human Activity Recognition (HAR) systems, using biceps concentration curls as an example. Findings show that fatigue prolongs completion time of later sets and decreases muscular endurance, leading to changes in data patterns and affecting the performance of subject-specific and cross-subject models. Feedforward Neural Network (FNN) exhibits the best performance in both types of models.
Article
Green & Sustainable Science & Technology
Gema Diaz-Quesada, Cecilia Bahamonde-Perez, Jose Maria Gimenez-Egido, Gema Torres-Luque
Summary: The study found that only 50% of children met the daily recommendations of 120 minutes of MVPA and 13,000 steps. There were no gender differences detected. Participants showed higher levels of PA and steps on weekends, and Out-of-School Time PA and steps were higher than School Time PA. This study suggests strategies and guidelines for physical activity in schools to promote well-being in early childhood.
Article
Computer Science, Information Systems
Ali Boudjema, Faiza Titouna, Chafiq Titouna
Summary: Human Activity Recognition (HAR) is an important research area driven by advancements in wearable device sensors. Traditional methods struggle with extracting relevant features from time series data, while deep learning models show promise in classification performance but face challenges in hyperparameter tuning, training, and complexity. This paper proposes AReNet, a lightweight deep learning model for HAR that achieves accurate activity recognition through a strategic approach.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Chemistry, Analytical
Aiguo Wang, Shenghui Zhao, Huan-Chao Keh, Guilin Chen, Diptendu Sinha Roy
Summary: This study proposes a clustering guided hierarchical framework for discriminating human activities. By introducing an activity confusion index based on clustering, the confusion between activities is quantitatively measured automatically, leading to the design of a hierarchical activity recognition framework to reduce recognition errors between similar activities.
Article
Sport Sciences
Anna Gabriela Silva Vilela Ribeiro, Alex Harley Crisp, Michele Novaes Ravelli, Maria Rita Marques de Oliveira, Rozangela Verlengia
Summary: The current study aimed to investigate the validity of three ActiGraph predictive equations for estimating free-living physical activity energy expenditure (PAEE) in women with severe obesity. The results showed that the Freedson VM3 Combination equation had higher precision at the group level, but all three equations had poor agreement and high uncertainty in estimating PAEE.
JOURNAL OF SPORTS SCIENCES
(2022)
Article
Biochemistry & Molecular Biology
Abbas Ali Mahdi, Jamal Akhtar Ansari, Priyanka Chaurasia, Mohammad Kaleem Ahmad, Shipra Kunwar, Sally McClean, Pratheepan Yogarajah
Summary: This study investigated the blood lead levels in pregnant women and umbilical cord blood, finding elevated lead levels in both groups. There was a strong positive correlation between the maternal and umbilical cord blood lead levels. Factors such as recent home painting and close proximity to traffic congestion were significantly associated with higher maternal blood lead levels. Education, mother age, fuel, and water sources did not show a significant association. Screening of blood lead levels and stricter regulations on lead-based products are recommended to reduce the exposure to lead in high-risk women and the general population.
INDIAN JOURNAL OF CLINICAL BIOCHEMISTRY
(2023)
Review
Computer Science, Artificial Intelligence
Sahraoui Dhelim, Liming Chen, Huansheng Ning, Chris Nugent
Summary: Advancements in AI offer promising opportunities for suicide risk assessment, but current research focuses primarily on depression and lacks studies and datasets related to suicidal behavior.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Xia Wang, Jun Liu, Chris Nugent, Ian Cleland, Yang Xu
Summary: The major challenge in mobile agent path planning is to determine an optimal control model and evaluate the control system's reliability in an uncertain environment. To address this challenge, a learning-verification integrated method is proposed, which includes a modified Q-learning algorithm to find the best Q-table, a probability model based on the agent's behavior, and the use of PRISM for automatic verification. A case study in a grids map demonstrates that the proposed method yields the largest expected reward.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Zhi Chen, Shuai Zhang, Sally McClean, Fionnuala Hart, Michael Milliken, Brahim Allan, Ian Kegel
Summary: This paper investigates the application of eye-tracking technologies in Human-Computer Interaction (HCI) research and proposes a method for integrating layout information from the BT Player with user gaze data to gain insights into user experience. The study also reveals several promising areas for future research, and the code used in the study is open access on GitHub.
Article
Computer Science, Theory & Methods
Sahraoui Dhelim, Liming Chen, Sajal K. Das, Huansheng Ning, Chris Nugent, Gerard Leavey, Dirk Pesch, Eleanor Bantry-White, Devin Burns
Summary: This article surveys the literature on social media analysis for detecting mental distress, with a focus on studies published since the COVID-19 outbreak. The authors propose new approaches to organizing and classifying the large amount of research in this emerging field, providing fresh insights and knowledge for interested communities. The article also discusses future research directions and niche areas in detecting mental health problems using social media data, as well as the technical, privacy, and ethical challenges in this rapidly growing field.
ACM COMPUTING SURVEYS
(2023)
Article
Health Care Sciences & Services
Weimin Ding, Shengli Wu, Chris Nugent
Summary: This paper proposes a multimodal fusion enabled ensemble approach for accurate recognition of human activities in a smart environment. Useful features from Bluetooth beacons, binary sensors, and smart floor are extracted using a fuzzy logic based-method and presented with variable-size temporal windows. A group of support vector machine classifiers are used for classification, and a weighted ensemble method with optimal weights obtained through the geometric framework is applied to obtain the final prediction. The proposed method is evaluated on the UJAmI dataset, and the experimental results demonstrate its efficacy and robustness.
HEALTH INFORMATICS JOURNAL
(2023)
Article
Engineering, Electrical & Electronic
Weiran Song, Hui Wang, Ultan F. Power, Enayetur Rahman, Judit Barabas, Jiandong Huang, James McLaughlin, Chris Nugent, Paul Maguire
Summary: This research investigates the potential of using low-cost portable near-infrared (NIR) spectroscopy and chemometrics to distinguish respiratory viruses. The results demonstrate the feasibility of this method for rapid, on-site, and low-cost virus prescreening for RSV and SeV, with the possibility of extending it to other viruses like SARS-CoV-2.
IEEE SENSORS JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Xia Wang, Jun Liu, Samuel J. Moore, Chris D. Nugent, Yang Xu
Summary: Smart homes offer convenience and assistance, and this paper proposes a hierarchical framework using Hidden Markov Model (HMM) for behavioural analysis. It suggests dividing the analysis into spatial transfer and sensor transfer layers. By integrating an implicit Markov model and probabilistic model checking, the composition and probability of occurrence of complex behaviour sequences are effectively analyzed.
INFORMATION FUSION
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Federico Cruciani, Paul McCullagh, Catherine Saunders, Colm Hayden, Claudia Chisari, Ian Cleland, Chris Nugent
Summary: Interdisciplinary research is recognized as an important tool in addressing societal challenges, and the global pandemic has accelerated the use of mobile technology for remote health interventions. This study presents a reusable software platform for implementing digital interventions through mobile applications. The platform consists of a secure server-side application for data storage, a customizable mobile app for data collection, and a customizable web portal for data analysis. Two pilot studies conducted during the pandemic demonstrated the platform's effectiveness in supporting remote interventions with positive feedback from participants.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Matias Garcia-Constantino, Alexandros Konios, Irvin Hussein Lopez-Nava, Pierre Pouliet, Idongesit Ekerete, Mustafa A. Mustafa, Chris Nugent, Gareth Morrison
Summary: This paper proposes a novel approach using accelerometer sensor data to identify personalised abnormal behaviour in Activities of Daily Living (ADLs), such as preparing and drinking tea or coffee. Abnormal behaviour in these activities can indicate a health problem or a hazardous incident, making it important to monitor ADLs for timely action. The approach considers individual user profiles and collects data from various sensors. Experimental results show that accelerometer data is sufficient to identify the main stages of the considered ADLs, suggesting that abnormal behaviour can be detected through signal and duration changes.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Rebecca Hand, Ian Cleland, Chris Nugent
Summary: This paper investigates the use of transfer learning with AlexNet to classify bed occupancy in temperature image data. Unlike traditional CNNs, which are generally developed and trained from scratch for low resolution thermal sensor image data, this study evaluates the use of pre-trained or fine-tuned CNNs. Three different fine-tuning configurations of the AlexNet architecture are evaluated and the best performing network achieves an accuracy of 0.973 on greyscale images with a temperature resolution of 220 x 220.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Rashid Kamal, Chris Nugent, Ian Cleland, Paul McCullagh
Summary: Catching the attention of users is commonly achieved through push notifications, but excessive notifications can disrupt users' tasks. To address this issue, a mobile app has been developed to collect user data and determine the optimal time for sending notifications.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Tian-Yu Ren, Long-Hao Yang, Chris Nugent, Fei-Fei Ye, Naomi Irvine, Jun Liu
Summary: With the aging population and rising healthcare costs, smart environments provide an effective and cost-efficient way to care for the elderly. This research focuses on sensor-based human activity recognition (HAR) in big data context and proposes a data-driven solution based on the extended belief rule base (EBRB) model. By using a new rule generation method, the research enhances the efficiency of the EBRB model and demonstrates its effectiveness through case studies.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Nicole E. Blackburn, Ian Cleland, Chris Nugent, Joseph G. McVeigh, Eilis M. McCaughan, Iseult M. Wilson
Summary: Breast reconstruction using the latissimus dorsi flap is a common management option for breast cancer patients after mastectomy. However, shoulder dysfunction caused by fear avoidance beliefs is often overlooked. This study found that fear avoidance beliefs were low in LD flap patients and there was no difference in shoulder kinematics between operated and non-operated shoulder, suggesting better outcomes for these patients.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Luis Cabanero, Alejandro Perez-Vereda, Chris Nugent, Ian Cleland, Ramon Hervas, Ivan Gonzalez
Summary: Digital twins are virtual representation of physical entities that allow monitoring without the constraints of real entities. This article proposes a meta-model for building a behavior model of an agent living in a sensorized house inside a simulator, motivated by the difficulties in ADLs research. The model follows a needs-driven approach and supports uncertainty in sensor data for generating realistic datasets. The tool has potential applications in smart environment design and ADLs research.
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON UBIQUITOUS COMPUTING & AMBIENT INTELLIGENCE (UCAMI 2022)
(2023)