Rafik Goubran
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
Research Interests
Rafik Goubran has a total of 89 co-authors. Here are the top co-authors by number of shared publications:
- Frank Knoefel
Memory Care Research Group, Bruyère Health Research Institute, Ottawa, ON, Canada
20 shared publications - Bruce Wallace
Carleton University, Ottawa, ON, Canada
12 shared publications - Pengcheng Xi
National Research Council Canada, Ottawa, Canada
8 shared publications - Brady Laska
Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada
7 shared publications - Boris Vukovic
Accessibility Institute, Carleton University, Ottawa, Canada
5 shared publications Bruce Wallace
5 shared publications
Latest Publications
- Open Access
Two-Step 3D Microphone Array Fusion Algorithm to Enhance Sound Source Location Measurements
This paper presents a novel two-step algorithm for microphone array fusion to enhance sound source localization (SSL) in three-dimensional indoor reverberant environments. Simulation analyses using simulated Room Impulse Responses (RIRs) reveal that Angle-of-Arrival (AoA) accuracy varies significantly with source position, causing certain microphone arrays to produce unreliable estimates. To mitigate this, the algorithm excludes microphone pairs with low-confidence AoAs, thereby improving overall localization accuracy. To extend the applicability of the approach, a generalized version of the algorithm is proposed for arbitrary room geometries and array positions on each wall. Its performance is assessed across three scenarios: (1) the original room geometry with arrays placed at the center of each wall; (2) a room with different dimensions; and (3) arrays placed at arbitrary positions on walls. The results show that the generalized algorithm achieves similar improvements as the original two-step method, approximately halving the localization error. Moreover, while room geometry and array placement influence SSL accuracy, the generalized method consistently reduces error across all cases. Three conventional AoA estimation methods are evaluated and their performance is compared in the baseline SSL. These findings highlight the robustness and practical value of the proposed algorithm on the baseline methods for improving SSL performance in acoustically challenging environments.
Open AccessTwo-Step 3D Microphone Array Fusion Algorithm to Enhance Sound Source Location Measurements
This paper presents a novel two-step algorithm for microphone array fusion to enhance sound source localization (SSL) in three-dimensional indoor reverberant environments. Simulation analyses using simulated Room Impulse Responses (RIRs) reveal that Angle-of-Arrival (AoA) accuracy varies significantly with source position, causing certain microphone arrays to produce unreliable estimates. To mitigate this, the algorithm excludes microphone pairs with low-confidence AoAs, thereby improving overall localization accuracy. To extend the applicability of the approach, a generalized version of the algorithm is proposed for arbitrary room geometries and array positions on each wall. Its performance is assessed across three scenarios: (1) the original room geometry with arrays placed at the center of each wall; (2) a room with different dimensions; and (3) arrays placed at arbitrary positions on walls. The results show that the generalized algorithm achieves similar improvements as the original two-step method, approximately halving the localization error. Moreover, while room geometry and array placement influence SSL accuracy, the generalized method consistently reduces error across all cases. Three conventional AoA estimation methods are evaluated and their performance is compared in the baseline SSL. These findings highlight the robustness and practical value of the proposed algorithm on the baseline methods for improving SSL performance in acoustically challenging environments.
- Open Access
- 8 December 2025
- Published by JMIR Publications Inc. in JMIR Research Protocols
- Vol. 14, e67628
Use of Head-Mounted Inertial Sensors for Mobility Tasks: Protocol for a Scoping Review
Background: Human activity recognition (HAR) is the use of technology to detect a person’s movements. Sensors can be passive like cameras monitoring an area, or active like those attached to a person’s watch. HAR is a burgeoning field whose uses span from personal health tracking to at-home physical degradation monitoring. The benefit of having sensors attached to a person for constant HAR tracking can be seen in the personal fitness field where people track their number of steps or distance ran. For older adults, HAR can be used in combination with mobility tasks like sit-to-stand, timed-up-and-go, and other tasks to monitor a person’s ability to be self-sufficient and to live alone at home. Together, HAR and mobility tasks are an effective method to measure health, especially in older adults. Consumer wearables like Fitbits and Apple watches are currently being used to measure many of these mobility tasks. The wide adoption of these devices is a result of their ease of use and simplicity. Despite this, most wearables go unused shortly after purchase. A potential solution is the use of consumer devices that are already adopted like headphones or hearing aids. These devices can perform the same measurements as the wearables but have the advantage of being used for other reasons than health monitoring. Despite this, no scoping review has summarized the current capabilities of what can be done using head-worn sensors. Objective: We aim to understand how head-worn devices have been used to monitor dynamic mobility tasks and if they have been compared to traditional methods (like smartwatches or in-sole sensors). Methods: This scoping review will be guided by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) framework. There is no time limit to when the papers must have been published. The databases included in the search are PubMed, Engineering Village, SCOPUS, and Web of Science. Using pre-established inclusion and exclusion criteria, 2 reviewers will screen each paper. We will follow 3 stages of the screening process: abstract and title screening, full-text screening, and full-text data extraction. Results: The search will be started once this scoping review protocol is published. Conclusions: Head-mounted devices have become part of everyday life and many already contain accelerometers or inertial measurement units. This scoping review will examine how these sensors have already been used in research to measure mobility tasks, and determine how consumer devices can be adapted to improve the lives of regular people. International Registered Report Identifier (IRRID): DERR1-10.2196/67628
- 8 December 2025
- Published by JMIR Publications Inc. in JMIR Research Protocols
- Vol. 14, e67628
Open AccessUse of Head-Mounted Inertial Sensors for Mobility Tasks: Protocol for a Scoping Review
Background: Human activity recognition (HAR) is the use of technology to detect a person’s movements. Sensors can be passive like cameras monitoring an area, or active like those attached to a person’s watch. HAR is a burgeoning field whose uses span from personal health tracking to at-home physical degradation monitoring. The benefit of having sensors attached to a person for constant HAR tracking can be seen in the personal fitness field where people track their number of steps or distance ran. For older adults, HAR can be used in combination with mobility tasks like sit-to-stand, timed-up-and-go, and other tasks to monitor a person’s ability to be self-sufficient and to live alone at home. Together, HAR and mobility tasks are an effective method to measure health, especially in older adults. Consumer wearables like Fitbits and Apple watches are currently being used to measure many of these mobility tasks. The wide adoption of these devices is a result of their ease of use and simplicity. Despite this, most wearables go unused shortly after purchase. A potential solution is the use of consumer devices that are already adopted like headphones or hearing aids. These devices can perform the same measurements as the wearables but have the advantage of being used for other reasons than health monitoring. Despite this, no scoping review has summarized the current capabilities of what can be done using head-worn sensors. Objective: We aim to understand how head-worn devices have been used to monitor dynamic mobility tasks and if they have been compared to traditional methods (like smartwatches or in-sole sensors). Methods: This scoping review will be guided by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) framework. There is no time limit to when the papers must have been published. The databases included in the search are PubMed, Engineering Village, SCOPUS, and Web of Science. Using pre-established inclusion and exclusion criteria, 2 reviewers will screen each paper. We will follow 3 stages of the screening process: abstract and title screening, full-text screening, and full-text data extraction. Results: The search will be started once this scoping review protocol is published. Conclusions: Head-mounted devices have become part of everyday life and many already contain accelerometers or inertial measurement units. This scoping review will examine how these sensors have already been used in research to measure mobility tasks, and determine how consumer devices can be adapted to improve the lives of regular people. International Registered Report Identifier (IRRID): DERR1-10.2196/67628
- 1 August 2025
- Published by Carleton University
Supportive Systems for Cognitive Health and Dementia : Survey of technology and the role of standards
This report examines technology-based products and services designed to support independent, healthy living, with the goal of identifying best practices in product development and features, uncovering gaps in current offerings, and providing guidance on the use of relevant standards and development needs for new standards or standards frameworks for products. The primary focus is on solutions that enable Aging-in-Place (AiP) for Persons Living with Dementia (PLWD), as this represents the most developed area within supportive living technologies. The products reviewed are a representative sample of those commercially available as of August 2023. The report includes summaries of two product analysis reviews.
- 1 August 2025
- Published by Carleton University
Supportive Systems for Cognitive Health and Dementia : Survey of technology and the role of standards
This report examines technology-based products and services designed to support independent, healthy living, with the goal of identifying best practices in product development and features, uncovering gaps in current offerings, and providing guidance on the use of relevant standards and development needs for new standards or standards frameworks for products. The primary focus is on solutions that enable Aging-in-Place (AiP) for Persons Living with Dementia (PLWD), as this represents the most developed area within supportive living technologies. The products reviewed are a representative sample of those commercially available as of August 2023. The report includes summaries of two product analysis reviews.
- Open Access
Sleep Stage Classification using Multimodal Embedding Fusion from EOG and PSM
Accurate sleep stage classification is essential for diagnosing sleep disorders, particularly in aging populations. While traditional polysomnography (PSG) relies on electroencephalography (EEG) as the gold standard, its complexity and need for specialized equipment make home-based sleep monitoring challenging. To address this limitation, we investigate the use of electrooculography (EOG) and pressure-sensitive mats (PSM) as less obtrusive alternatives for five-stage sleep-wake classification. This study introduces a novel approach that leverages ImageBind, a multimodal embedding deep learning model, to integrate PSM data with dual-channel EOG signals for sleep stage classification. Our method is the first reported approach that fuses PSM and EOG data for sleep stage classification with ImageBind. Our results demonstrate that fine-tuning ImageBind significantly improves classification accuracy, outperforming existing models based on single-channel EOG (DeepSleepNet), exclusively PSM data (ViViT), and other multimodal deep learning approaches (MBT). Notably, the model also achieved strong performance without fine-tuning, highlighting its adaptability to specific tasks with limited labeled data, making it particularly advantageous for medical applications. We evaluated our method using 85 nights of patient recordings from a sleep clinic. Our findings suggest that pre-trained multimodal embedding models, even those originally developed for non-medical domains, can be effectively adapted for sleep staging, with accuracies approaching systems that require complex EEG data.
Open AccessSleep Stage Classification using Multimodal Embedding Fusion from EOG and PSM
Accurate sleep stage classification is essential for diagnosing sleep disorders, particularly in aging populations. While traditional polysomnography (PSG) relies on electroencephalography (EEG) as the gold standard, its complexity and need for specialized equipment make home-based sleep monitoring challenging. To address this limitation, we investigate the use of electrooculography (EOG) and pressure-sensitive mats (PSM) as less obtrusive alternatives for five-stage sleep-wake classification. This study introduces a novel approach that leverages ImageBind, a multimodal embedding deep learning model, to integrate PSM data with dual-channel EOG signals for sleep stage classification. Our method is the first reported approach that fuses PSM and EOG data for sleep stage classification with ImageBind. Our results demonstrate that fine-tuning ImageBind significantly improves classification accuracy, outperforming existing models based on single-channel EOG (DeepSleepNet), exclusively PSM data (ViViT), and other multimodal deep learning approaches (MBT). Notably, the model also achieved strong performance without fine-tuning, highlighting its adaptability to specific tasks with limited labeled data, making it particularly advantageous for medical applications. We evaluated our method using 85 nights of patient recordings from a sleep clinic. Our findings suggest that pre-trained multimodal embedding models, even those originally developed for non-medical domains, can be effectively adapted for sleep staging, with accuracies approaching systems that require complex EEG data.
- 6 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-10
Coughprint: Distilled Cough Representations From Speech Foundation Model Embeddings
Supportive smart home systems with integrated sensors capable of measuring cough frequency and severity can support independent living and aging in place by helping monitor the state of acute and chronic health conditions. Previously, we showed that embeddings from speech foundation models are effective cough representations for a range of cough measurement applications. While powerful, the large compute and memory requirements of these models prevents them from being deployed in embedded smart sensors. In this work we use knowledge distillation to train edge-compute focused student models, making them feasible for the measurement, identification, and classification of cough sounds directly in the smart home. This embedded processing avoids the privacy and security concerns associated with transmission and storage of sensitive audio recordings in the cloud. We show that the student networks preserve the universal cough representation capabilities of the teacher, even generalizing to unseen classes such as speech, allowing the same network to be used for multiple downstream applications without any task-specific fine-tuning. A student network based on a 14-layer variant of ResNet achieved the highest aggregate quality score across the downstream evaluation tasks, even outperforming the foundation model teacher on certain tasks despite having over 200× fewer parameters. Linear classification on the embeddings from the proposed student network achieves strong performance on a diverse set of cough measurement tasks, scoring 98.3% on cough/non-cough discrimination, 90.3% on human sound classification, 94.8% on cougher verification, 84.4% on cougher identification, and 87.8% on wet/dry cough classification.
- 6 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-10
Coughprint: Distilled Cough Representations From Speech Foundation Model Embeddings
Supportive smart home systems with integrated sensors capable of measuring cough frequency and severity can support independent living and aging in place by helping monitor the state of acute and chronic health conditions. Previously, we showed that embeddings from speech foundation models are effective cough representations for a range of cough measurement applications. While powerful, the large compute and memory requirements of these models prevents them from being deployed in embedded smart sensors. In this work we use knowledge distillation to train edge-compute focused student models, making them feasible for the measurement, identification, and classification of cough sounds directly in the smart home. This embedded processing avoids the privacy and security concerns associated with transmission and storage of sensitive audio recordings in the cloud. We show that the student networks preserve the universal cough representation capabilities of the teacher, even generalizing to unseen classes such as speech, allowing the same network to be used for multiple downstream applications without any task-specific fine-tuning. A student network based on a 14-layer variant of ResNet achieved the highest aggregate quality score across the downstream evaluation tasks, even outperforming the foundation model teacher on certain tasks despite having over 200× fewer parameters. Linear classification on the embeddings from the proposed student network achieves strong performance on a diverse set of cough measurement tasks, scoring 98.3% on cough/non-cough discrimination, 90.3% on human sound classification, 94.8% on cougher verification, 84.4% on cougher identification, and 87.8% on wet/dry cough classification.
- Open Access
Sleep Position Classification using Transfer Learning for Bed-based Pressure Sensors
Bed-based pressure-sensitive mats (PSMs) offer a non-intrusive way of monitoring patients during sleep. We focus on four-way sleep position classification using data collected from a PSM placed under a mattress in a sleep clinic. Sleep positions can affect sleep quality and the prevalence of sleep disorders, such as apnea. Measurements were performed on patients with suspected sleep disorders referred for assessments at a sleep clinic. Training deep learning models can be challenging in clinical settings due to the need for large amounts of labeled data. To overcome the shortage of labeled training data, we utilize transfer learning to adapt pre-trained deep learning models to accurately estimate sleep positions from a low-resolution PSM dataset collected in a polysomnography sleep lab. Our approach leverages Vision Transformer models pre-trained on ImageNet using masked autoencoding (ViTMAE) and a pre-trained model for human pose estimation (ViTPose). These approaches outperform previous work from PSM-based sleep pose classification using deep learning (TCN) as well as traditional machine learning models (SVM, XGBoost, Random Forest) that use engineered features. We evaluate the performance of sleep position classification from 112 nights of patient recordings and validate it on a higher resolution 13-patient dataset. Despite the challenges of differentiating between sleep positions from low-resolution PSM data, our approach shows promise for real-world deployment in clinical settings
Open AccessSleep Position Classification using Transfer Learning for Bed-based Pressure Sensors
Bed-based pressure-sensitive mats (PSMs) offer a non-intrusive way of monitoring patients during sleep. We focus on four-way sleep position classification using data collected from a PSM placed under a mattress in a sleep clinic. Sleep positions can affect sleep quality and the prevalence of sleep disorders, such as apnea. Measurements were performed on patients with suspected sleep disorders referred for assessments at a sleep clinic. Training deep learning models can be challenging in clinical settings due to the need for large amounts of labeled data. To overcome the shortage of labeled training data, we utilize transfer learning to adapt pre-trained deep learning models to accurately estimate sleep positions from a low-resolution PSM dataset collected in a polysomnography sleep lab. Our approach leverages Vision Transformer models pre-trained on ImageNet using masked autoencoding (ViTMAE) and a pre-trained model for human pose estimation (ViTPose). These approaches outperform previous work from PSM-based sleep pose classification using deep learning (TCN) as well as traditional machine learning models (SVM, XGBoost, Random Forest) that use engineered features. We evaluate the performance of sleep position classification from 112 nights of patient recordings and validate it on a higher resolution 13-patient dataset. Despite the challenges of differentiating between sleep positions from low-resolution PSM data, our approach shows promise for real-world deployment in clinical settings
- Open Access
- 24 October 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Use of Head-Mounted Inertial Sensors for Mobility Tasks: Protocol for a Scoping Review (Preprint)
BACKGROUND Human activity recognition (HAR) is the use of technology to detect a person’s movements. Sensors can be passive like cameras monitoring an area, or active like those attached to a person’s watch. HAR is a burgeoning field whose uses span from personal health tracking to at-home physical degradation monitoring. The benefit of having sensors attached to a person for constant HAR tracking can be seen in the personal fitness field where people track their number of steps or distance ran. For older adults, HAR can be used in combination with mobility tasks like sit-to-stand, timed-up-and-go, and other tasks to monitor a person’s ability to be self-sufficient and to live alone at home. Together, HAR and mobility tasks are an effective method to measure health, especially in older adults. Consumer wearables like Fitbits and Apple watches are currently being used to measure many of these mobility tasks. The wide adoption of these devices is a result of their ease of use and simplicity. Despite this, most wearables go unused shortly after purchase. A potential solution is the use of consumer devices that are already adopted like headphones or hearing aids. These devices can perform the same measurements as the wearables but have the advantage of being used for other reasons than health monitoring. Despite this, no scoping review has summarized the current capabilities of what can be done using head-worn sensors. OBJECTIVE We aim to understand how head-worn devices have been used to monitor dynamic mobility tasks and if they have been compared to traditional methods (like smartwatches or in-sole sensors). METHODS This scoping review will be guided by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) framework. There is no time limit to when the papers must have been published. The databases included in the search are PubMed, Engineering Village, SCOPUS, and Web of Science. Using pre-established inclusion and exclusion criteria, 2 reviewers will screen each paper. We will follow 3 stages of the screening process: abstract and title screening, full-text screening, and full-text data extraction. RESULTS The search will be started once this scoping review protocol is published. CONCLUSIONS Head-mounted devices have become part of everyday life and many already contain accelerometers or inertial measurement units. This scoping review will examine how these sensors have already been used in research to measure mobility tasks, and determine how consumer devices can be adapted to improve the lives of regular people. INTERNATIONAL REGISTERED REPORT DERR1-10.2196/67628
- 24 October 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessUse of Head-Mounted Inertial Sensors for Mobility Tasks: Protocol for a Scoping Review (Preprint)
BACKGROUND Human activity recognition (HAR) is the use of technology to detect a person’s movements. Sensors can be passive like cameras monitoring an area, or active like those attached to a person’s watch. HAR is a burgeoning field whose uses span from personal health tracking to at-home physical degradation monitoring. The benefit of having sensors attached to a person for constant HAR tracking can be seen in the personal fitness field where people track their number of steps or distance ran. For older adults, HAR can be used in combination with mobility tasks like sit-to-stand, timed-up-and-go, and other tasks to monitor a person’s ability to be self-sufficient and to live alone at home. Together, HAR and mobility tasks are an effective method to measure health, especially in older adults. Consumer wearables like Fitbits and Apple watches are currently being used to measure many of these mobility tasks. The wide adoption of these devices is a result of their ease of use and simplicity. Despite this, most wearables go unused shortly after purchase. A potential solution is the use of consumer devices that are already adopted like headphones or hearing aids. These devices can perform the same measurements as the wearables but have the advantage of being used for other reasons than health monitoring. Despite this, no scoping review has summarized the current capabilities of what can be done using head-worn sensors. OBJECTIVE We aim to understand how head-worn devices have been used to monitor dynamic mobility tasks and if they have been compared to traditional methods (like smartwatches or in-sole sensors). METHODS This scoping review will be guided by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) framework. There is no time limit to when the papers must have been published. The databases included in the search are PubMed, Engineering Village, SCOPUS, and Web of Science. Using pre-established inclusion and exclusion criteria, 2 reviewers will screen each paper. We will follow 3 stages of the screening process: abstract and title screening, full-text screening, and full-text data extraction. RESULTS The search will be started once this scoping review protocol is published. CONCLUSIONS Head-mounted devices have become part of everyday life and many already contain accelerometers or inertial measurement units. This scoping review will examine how these sensors have already been used in research to measure mobility tasks, and determine how consumer devices can be adapted to improve the lives of regular people. INTERNATIONAL REGISTERED REPORT DERR1-10.2196/67628
- Open Access
Respiratory Rate Estimation from Thermal Video Data Using Spatio-Temporal Deep Learning
Thermal videos provide a privacy-preserving yet information-rich data source for remote health monitoring, especially for respiration rate (RR) estimation. This paper introduces an end-to-end deep learning approach to RR measurement using thermal video data. A detection transformer (DeTr) first finds the subject’s facial region of interest in each thermal frame. A respiratory signal is estimated from a dynamically cropped thermal video using 3D convolutional neural networks and bi-directional long short-term memory stages. To account for the expected phase shift between the respiration measured using a respiratory effort belt vs. a facial video, a novel loss function based on negative maximum cross-correlation and absolute frequency peak difference was introduced. Thermal recordings from 22 subjects, with simultaneous gold standard respiratory effort measurements, were studied while sitting or standing, both with and without a face mask. The RR estimation results showed that our proposed method outperformed existing models, achieving an error of only 1.6 breaths per minute across the four conditions. The proposed method sets a new State-of-the-Art for RR estimation accuracy, while still permitting real-time RR estimation.
Open AccessRespiratory Rate Estimation from Thermal Video Data Using Spatio-Temporal Deep Learning
Thermal videos provide a privacy-preserving yet information-rich data source for remote health monitoring, especially for respiration rate (RR) estimation. This paper introduces an end-to-end deep learning approach to RR measurement using thermal video data. A detection transformer (DeTr) first finds the subject’s facial region of interest in each thermal frame. A respiratory signal is estimated from a dynamically cropped thermal video using 3D convolutional neural networks and bi-directional long short-term memory stages. To account for the expected phase shift between the respiration measured using a respiratory effort belt vs. a facial video, a novel loss function based on negative maximum cross-correlation and absolute frequency peak difference was introduced. Thermal recordings from 22 subjects, with simultaneous gold standard respiratory effort measurements, were studied while sitting or standing, both with and without a face mask. The RR estimation results showed that our proposed method outperformed existing models, achieving an error of only 1.6 breaths per minute across the four conditions. The proposed method sets a new State-of-the-Art for RR estimation accuracy, while still permitting real-time RR estimation.