Publications

  Authors: Hawkar Oagaz, Breawn Schoun, Manpreet Pooji, Min-Hyung Choi

  Publication:IEEE Journal of Biomedical and Health Informatics

  Category: Journal

  Link: @IEEE

Abstract:The ability to detect and diagnose neurocognitive disorders at the earliest possible moment is key to a better prognosis for the patient. Two of the earliest indicators of potential neurocognitive problems are motor and visual dysfunction. Motor disorders and problems in visual cognition can be seen in many neurocognitive disorders, resulting in abnormal physical reactions to visual stimuli. Analyzing physical behaviors when presented with such stimuli can provide insights into the visual perception and motor abilities of an individual, yet there is currently no unbiased, objective, general-purpose tool that analyzes attention and motor behavior to assess neurocognitive function. We propose a novel method of neurocognitive function assessment that tests the patient's cognition using Virtual Reality (VR) with eye tracking and motion analysis. By placing the patient in a controlled virtual environment and analyzing their movements, we can evoke certain physical responses from subjects for neurocognitive assessment. We have developed a prototype system that places the subject in a virtual baseball field and captures their full body motion as they try to catch baseballs. This scenario tests the subject's ability to determine the landing time and position of the ball, as well as the test subject's balance, motor skills, attention, and memory. Preliminary tests with 20 healthy normal individuals demonstrate the ability of this tool to assess the test subject's balance, memory, attention, and reaction to visual stimuli. This platform has a twofold contribution: it is used to assess several neurocognitive constructs that affect visual and motor capability neutrally and objectively based on controlled stimuli, and it enables objective comparison between different neurocognitive disorders research in this field.

  Authors: Abeer Abdel Khaleq, Ilkyeun Ra

  Publication:The 10th International conference on the Internet (ICONI) 2018

  Category: Conference / Oral paper presentation

  Link: TBA

Abstract:Disaster management requires a system architecture that satisfies an agile, dynamic, reliable, and scalable solution. When a disaster strikes, relevant information needs to be extracted and delivered on time to decision makers to provide the needed resources through all disaster phases of preparedness, response and recovery. This paper focuses on the big data aspect of disaster management using a microservice approach hosted on the cloud. Data flow through the system is orchestrated through a hybrid approach of data analytics jobs and microservices implemented on the Microsoft Azure platform. Twitter data analytics from hurricane disasters is used as the source for big data flow providing both historical and real time disaster data processing. Initial implementation of the system shows that a microservice approach delivers a system that can handle the aspects of big data with more agility, scalability and reliability suited for a disaster management system as a service

  Authors: Nguyen, Anh and Alqurashi, Raghda and Raghebi, Zohreh and Banaei-Kashani, Farnoush and Halbower, Ann C. and Vu, Tam

  Publication:Commun. ACM

  Category: Journal

  Link: @ACM

Abstract:Sensing physiological signals from the human head has long been used for medical diagnosis, human-computer interaction, meditation quality monitoring, among others. However, existing sensing techniques are cumbersome and not desirable for long-term studies and impractical for daily use. Due to these limitations, we explore a new form of wearable systems, called LIBS, that can continuously record biosignals such as brain wave, eye movements, and facial muscle contractions, with high sensitivity and reliability. Specifically, instead of placing numerous electrodes around the head, LIBS uses a minimal number of custom-built electrodes to record the biosignals from human ear canals. This recording is a combination of three signals of interest and unwanted noise. Therefore, we design an algorithm using a supervised Nonnegative Matrix Factorization (NMF) model to split the single-channel mixed signal into three individual signals representing electrical brain activities (EEG), eye movements (EOG), and muscle contractions (EMG). Through prototyping and implementation over a 30 day sleep experiment conducted on eight participants, our results prove the feasibility of concurrently extracting separated brain, eye, and muscle signals for fine-grained sleep staging with more than 95% accuracy. With this ability to separate the three biosignals without loss of their physiological information, LIBS has a potential to become a fundamental in-ear biosensing technology solving problems ranging from self-caring health to non-health and enabling a new form of human communication interfaces.

  Authors: Raghebi, Zohreh and Banaei-Kashani, Farnoush

  Publication:A2018 IEEE International Conference on Big Data (Big Data)

  Category: Conference

  Link: @IEEE

Abstract:With the rapid development of location sensors, it is now possible to accurately study how various items (such as viruses or messages) spread across populations of moving objects. In such applications, an item can propagate through the object population where two objects are close. Such a dynamic network of objects called a "contact network". In this paper, we define and study a family of probabilistic reachability queries in uncertain contact networks, where contacts between objects are probabilistic. A probabilistic reachability query verifies whether two objects are "reachable" with a probability no less than a threshold η. To enable efficient processing of probabilistic reachability queries on large uncertain contact networks, first, we present a series-parallel reduction technique that significantly reduces the size of the input uncertain contact network in order to shrink the search space while maintaining accuracy and second, we introduce Optimized Spatiotemporal Tree Cover, an index structure that leverages the spatiotemporal properties of the contact network. With an extensive analytical and empirical study, we demonstrate superiority of our proposed solution versus a baseline solution (i.e., Monte Carlo sampling) and the only other existing solution with 400% and 200% improvement in query processing time on average, respectively.

  Authors: Raghebi, Zohreh and Banaei-Kashani, Farnoush

  Publication:ACM GeoRich '18

  Category: Workshop

  Link: @ACM

Abstract:With the advent of reliable positioning technologies and prevalence of location-based services, it is now feasible to accurately study the propagation of items such as infectious viruses, sensitive information pieces, and malwares through a population of moving objects, e.g., individuals, vehicles, and mobile devices. In such application scenarios, an item passes between two objects when the objects are sufficiently close (i.e., when they are, so-called, in contact), and hence once an item is initiated, it can propagate in the object population through the evolving network of contacts among objects, termed contact network. In this paper, for the first time we define and study probabilistic reachability queries in large uncertain contact networks, where propagation of items through contacts are uncertain. A probabilistic reachability query verifies whether two objects are "reachable" through the evolving uncertain contact network with a probability greater than a threshold η. For efficient processing of probabilistic queries, we propose a novel index structure, termed spatiotemporal tree cover (STC), which leverages the spatiotemporal properties of the contact network for efficient processing of the queries. Our experiments with real data demonstrate superiority of our proposed solution versus the only other existing solution (based on Monte Carlo sampling) for processing probabilistic reachability queries in generic uncertain graphs, with 300% improvement in query processing time on average.

  Authors: Raghebi, Zohreh and Banaei-Kashani, Farnoush

  Publication:ACM SIGSPATIAL '18

  Category: Conference

  Link: @ACM

Abstract:With the rapid development of location sensors, it is now possible to study how various items (such as viruses and messages) spread across populations of moving objects at scale. In such applications, two objects are considered in-contact while they are sufficiently close to each other. Such a dynamic network of objects, so-called a "contact network". In this paper, we define and study probabilistic reachability queries in uncertain contact networks, where contacts between objects are probabilistic. A probabilistic reachability query verifies whether two objects are "reachable" with a probability no less than a threshold η. We introduce Optimized Spatiotemporal Tree Cover, an index structure that leverages the spatiotemporal properties of the contact network to enable efficient processing of the reachability queries on large uncertain contact networks. With an extensive study using both real and synthetic datasets, we demonstrate superiority of our proposed solution versus a baseline solution (i.e., Monte Carlo Sampling) and the only other existing solution for reachability queries on uncertain contact networks, with 350% and 150% improvement in query processing time on average, respectively.

  Authors: Joel Fredrickson, Omar Alqahtani, Farnoush Banaei-Kashani, Michael Mannino

  Publication: WITS 2018

  Category: Conference

  Link: TBA

Abstract:We extend a similarity measure for medical event sequences (MESs) with linked patient records and evaluate its classification performance for retrospective mortality prediction among trauma patients, a benchmark prediction task in medical decision-making. We extend the Optimal Temporal Common Subsequence for MESs (OTCS-MES) measure with distance functions and weights to combine key variables of patient records with MES similarity. Our empirical evaluation compares the predictive performance of the Trauma Mortality Prediction Model (TMPM), an accepted regression approach for mortality prediction in trauma data, to nearest neighbor algorithms using similarity measures based on medical event history and linked patient records. Using a large data set of trauma incidents from the National Trauma Data Bank, our results indicate improved predictive performance for an ensemble of nearest neighbor classifiers over TMPM augmented with a second stage regression using patient variables. Furthermore, when supplementing our similarity measure with patient attributes, we see improved predictive performance over measures based solely on medical event sequences. Results provide additional evidence that similarity measures for medical event sequences are a powerful and easily adapted method for medical decision-making.

  Authors: Abeer Abdel Khaleq, Ilkyeun Ra

  Publication: IEEE Global Humanitarian Technology Conference 2018

  Category: Conference

  Link: @IEEE

Abstract:Disasters whether natural or man-made have great impact on countries and civilians. Proper information across the main disaster phases need to be delivered on time and to the right people to minimize the impact and provide needed resources. Social media and Twitter in particular, is an important mean of information sharing in real-time as part of a complete cyber-physical emergency management system during a disaster. Twitter can be used in any place in the world through smartphones or other mediums with an internet access connection. The vast and varied number of tweets produced during a disaster will benefit from the cloud scalable storage and processing resources. As a centralized processing system is more vulnerable when a disaster strikes, there is a need for a more resilient distributed system architecture that allows for the distribution of both processing and storage resources. The goal of our study is to develop and evaluate a prototype of a microservice architecture for twitter data analytics during a disaster that meets the requirements of disaster management. In this paper, we design a cloud-based microservices twitter analytics framework for disaster management and implement a basic prototype system. Our prototype system demonstrates that the microservices approach allows for a distributed, dynamic, reliable and scalable system architecture on cloud platform that goes in hand with disaster domain requirements.

  Authors: Abeer Abdel Khaleq, Ilkyeun Ra

  Publication: IEEE Future Technologies Conference 2018

  Category: Conference

  Link: @Springer

Abstract:Natural disasters happen at any time and at any place. Social media can provide an important mean for both people affected and emergency personnel in sharing and receiving relevant information as the disaster unfolds across the different phases of the disaster. Focusing on the phases of preparedness, response and recovery, certain information needs to be retrieved due to the critical mission of emergency personnel. Such information can be directed depending on the disaster phase towards warning citizens, saving lives, or reducing the disaster impact. In this paper, we present an analytical study on Twitter data for three recent major hurricane disasters covering the three main disaster phases of preparedness, response and recovery. Our goal is to identify relevant tweets that will carry important information for disaster phase discovery. To achieve our goal, we propose a cloud-based system framework focused on three main components of disaster relevance classification, disaster phase classification and knowledge extraction. The framework is general enough for the three main disaster phases and specific to a hurricane disaster. Our results show that relevant tweets from different disaster data sets spanning different disaster phases can be classified for relevancy with an accuracy around 0.86, and for disaster phase with an accuracy of 0.85, where key information for disaster management personnel can be extracted.

  Authors: Joel Fredrickson, Omar Alqahtani, Michael Mannino, Farnoush Banaei-Kashani

  Publication: AIS Americas Conference on Information Systems 2018

  Category: Conference

  Link: @AIS

Abstract:We extend a similarity measure for medical event sequences (MESs) and evaluate its performance on mortality prediction using a substantial trauma data set. We extend the Optimal Temporal Common Subsequence for MESs (OTCS-MES) measure by generalizing the event-matching component with user- defined weights. In the empirical evaluation of classification performance, we provide a more complete evaluation than previous studies. We compare the predictive performance of the Trauma Mortality Prediction Model (TMPM), an accepted regression approach for mortality prediction in trauma data, to nearest neighbor algorithms using similarity measures for MESs. Using a data set from the National Trauma Data Bank, our results indicate improved predictive performance for an ensemble of nearest neighbor classifiers over TMPM. Our analysis demonstrates a superior Receiver Operating Characteristics (ROC) curve, larger AUC, and improved operating points on a ROC curve. Predictive performance improves for the ensemble for a variety of sensitivity weights and false positive constraints.

  Authors: Breawn Schoun, Shane Transue, Ann C. Halbower, Min-Hyung Choi

  Publication: IEEE Conference on Biomedical and Health Informatics 2018

  Category: Conference

  Acceptance Rate: 38%

  Link: @GraphicsLab

Abstract:Respiration monitoring methods that are both accurate and comfortable are highly sought after in the medical field. No existing method of respiration monitoring perfectly satisfies both of these criteria; each method is a trade-off between comfort and accuracy. Contact methods, which require placing sensors directly on the patient’s body, provide reliable measurements, but are uncomfortable for the patient, which alters their natural breathing behaviors. Conversely, non- contact methods monitor respiration remotely and comfortably, but with lower accuracy. We present a method of respiratory analysis that is non-contact, but also measures the exhaled air of a human subject directly through a medium-based exhale visualization technique. In this method, we place a thin medium perpendicular to the exhaled airflow of an individual, and use a thermal camera to record the heat signature from the exhaled breath on the opposite side of the material. Breathing rate and respiratory behaviors are extracted from the thermal data in real time. Our proposed respiration monitoring technique ac- curately reports breathing rate, and provides other information not obtainable through other non-contact methods. This method can be implemented as a small low-cost device for ease of use in a clinical environment.

2018 - Behavioral Analysis of Turbulent Exhale Flows

  Authors:Shane Transue, Sayed Mohsin Reza, Ann C. Halbower, Min-Hyung Choi

  Publication:IEEE Conference on Biomedical and Health Informatics 2018

  Category: Conference

  Acceptance Rate: 38%

  Link: @GraphicsLab

Abstract:Dense exhale flow through CO2 spectral imaging introduces a pivotal trajectory within non-contact respiratory analysis that consolidates several pulmonary evaluations into a single coherent monitoring process. Due to technical limitations and the limited exploration of respiratory analysis through this non-contact technique, this method has not been fully utilized to extract high-level respiratory behaviors through turbulent exhale analysis. In this work, we present a structural foundation for respiratory analysis of turbulent exhale flows through the visualization of dense CO2 density distributions using pre- cisely refined thermal imaging device to target high-resolution respiratory modeling. We achieve spatial and temporal high- resolution flow reconstructions through the cooperative devel- opment of a thermal camera dedicated to respiratory analysis to drastically improve the precision of current exhale imaging methods. We then model turbulent exhale behaviors using a heuristic volumetric flow reconstruction process to generate sparse flow exhale models. Together these contributions allow us to target the acquisition of numerous respiratory behaviors including, breathing rate, exhale strength and capacity, towards insights into lung functionality and tidal volume estimation.

  Authors: Breawn Schoun, Shane Transue, Min-Hyung Choi

  Publication:International Conference for High Performance Computing, Networking, Storage and Analysis 2017

  Category: Conference

  Acceptance Rate: 18%

  Link: @GraphicsLab

Abstract:Respiration monitoring is an important physiological measurement taken to determine the health of an individual. In clinical sleep studies, respiration activity is monitored to detect sleep disorders such as sleep apnea and respiratory conditions such as Chronic Obstructive Pulmonary Disease (COPD). Existing methods of respiration monitoring either place sensors on the patient's body, causing discomfort to the patient, or monitor respiration remotely with lower accuracy. We present a method of respiratory analysis that is non-contact, but also measures the exhaled air of a human subject directly through a medium-based exhale visualization technique. In this method, we place a thin medium perpendicular to the exhaled airflow of an individual, and use a thermal camera to record the heat signature from the exhaled breath on the opposite side of the material. Respiratory behaviors are extracted from the thermal data in real time using Python. Our prototype is an embedded, low-power device that performs image and signal processing in realtime with Python, making use of powerful existing Python modules for scientific computing and visualization. Our proposed respiration monitoring technique accurately reports breathing rate, and may provide other metrics not obtainable through other non-contact methods. This method can be useful for medical applications where long-term respiratory analysis is necessary, and for applications that require additional information about breathing behavior.

  Authors:Mannino, M., Fredrickson, J., Banaei-Kashani, F., Linck, I., & Raghda, R. A

  Publication:ACM Transactions on Management Information Systems (TMIS)

  Category: Journal

  Acceptance Rate: 33%

  Link: @ACM

Abstract:We develop a similarity measure for medical event sequences (MESs) and empirically evaluate it using U.S. Medicare claims data. Existing similarity measures do not use unique characteristics of MESs and have never been evaluated on real MESs. Our similarity measure, the Optimal Temporal Common Subsequence for Medical Event Sequences (OTCS-MES) provides a matching component that integrates event prevalence, event duplication, and hierarchical coding, important elements of MESs. The OTCS-MES also uses normalization to mitigate the impact of heavy positive skew of matching events and compact distribution of event prevalence. We empirically evaluate the OTCS-MES measure against two other measures specifically designed for MESs, the original OTCS and Artemis, a measure incorporating event alignment. Our evaluation uses two substantial data sets of Medicare claims data containing inpatient and outpatient sequences with different medical event coding. We find a small overlap in nearest neighbors among the three similarity measures, demonstrating the superior design of the OTCS-MES with its emphasis on unique aspects of MESs. The evaluation also provides evidence about the impact of component weights, neighborhood size, and sequence length.

  Authors: Breawn Schoun, Shane Transue, Min-Hyung Choi

  Publication: Center of Excellence in Wireless and Information Technology Conference 2017

  Category: Conference

  Link: @GraphicsLab

Abstract:Respiration monitoring is an important physiological measurement taken to determine numerous health attributes of an individual. In clinical sleep studies, respiration rate is monitored to detect sleep disorders such as sleep apnea and respiratory conditions such as Chronic Obstructive Pulmonary Disease (COPD). Methods of respiration monitoring fall into two categories: contact methods, which monitor respiration by placing sensors on the patient’s body, and non-contact methods, which monitor respiration without direct contact with the patient through remote sensors. Contact methods maintain higher accuracies than non-contact methods because they monitor respiration directly, but these methods often cause discomfort to the patient and alter natural breathing behaviors. One such device is the spirometer, which requires the patient to wear a clip on their nose and breathe forcefully through a tube. Not only is this method uncomfortable and strenuous, it also requires a conscious decision on the part of the patient to breathe in a particular manner, making it unfit for use with sleeping patients. Due to these factors, this device and other contact methods can only be used for a short period of time, making them unsuitable for long-term studies. Non-contact methods have an opposite tradeoff; while they are more comfortable and preserve natural breathing, they measure respiration indirectly, and therefore less accurately. To exploit this tradeoff for both comfort and natural respiration monitoring, we present a novel method of non-contact respiratory analysis that improves on current methods by measuring the exhaled air of a human subject through a medium-based exhale visualization technique. In this method, we place a thin medium perpendicular to the exhaled airflow of an individual, and use a low-cost thermal camera to record the heat signature from the exhaled breath on the opposite side of the material. Breathing rate and respiratory behaviors are then extracted from the thermal data using image processing techniques. Data collected from our experiments shows strong correlations between the exhale behaviors and the heat signature within the medium. As a proof of concept, we simulate exhaled air by using a fan to blow heated air onto a medium to validate the airflow to thermal imaging relation. We also test this technique on several human subjects to explore clinical feasibility. Like other respiration monitoring methods, this technique accurately reports breathing rate, but also provides metrics not obtainable through other non-contact methods, such as breathing strength, nose to mouth distribution, and tidal volume estimates. Nose to mouth distribution behavior is a metric not currently attainable from either contact or non-contact methods, and because nasal obstruction is known to increase the risk of sleep-disordered breathing problems, this measurement is highly valuable in sleep studies. This method can also be useful for a variety of medical applications where long-term respiratory analysis is necessary, and is particularly useful for applications that require additional information about breathing behavior.

  Authors:Shane Transue, Phuc Nguyen, Tam Vu, Min-Hyung Choi

  Publication:Conference on Connected Health: Applications, Systems, and Engineering (CHASE) 2017

  Category: Conference

  Link: @IEEE

Abstract:Reliable occluded skeletal posture estimation is a fundamentally challenging problem for vision-based monitoring techniques. This is due to several imaging related challenges introduced by existing depth-based pose estimation techniques that fail to provide accurate joint position estimates when the line of sight between the imaging device and the patient is obscured by an occluding material. In this work, we present a new method of estimating skeletal posture in occluded applications using both depth and thermal imaging through volumetric modeling and introduce a new occluded ground-truth tracking method inspired by modern motion capture solutions. Using this integrated volumetric model, we utilize Convolutional Neural Networks to characterize and identify volumetric thermal distributions that match trained skeletal posture estimates which includes disconnected skeletal definitions and allows correct posture estimation in highly ambiguous cases. We demonstrate this approach by correctly identifying common sleep postures that present challenging cases for current skeletal joint estimations, obtaining an average classification accuracy of ~94.45%.

  Authors:Shane Transue, Phuc Nguyen, Tam Vu, Min-Hyung Choi

  Publication:Conference on Connected Health: Applications, Systems, and Engineering (CHASE) 2016

  Category: Conference

  Link: @IEEE

Abstract:Breathing volume measurement has long been an important physiological indication widely used for the diagnosis and treatment of pulmonary diseases. However, most of existing breathing volume monitoring techniques require either physical contact with the patient or are prohibitively expensive. In this paper we present an automated and inexpensive non-contact, vision-based method for monitoring an individual's tidal volume, which is extracted from a three-dimensional (3D) chest surface reconstruction from a single depth camera. In particular, formulating the respiration monitoring process as a 3D space-time volumetric representation, we introduce a real-time surface reconstruction algorithm to generate omni-direction deformation states of a patient's chest while breathing, which reflects the change in tidal volume over time. These deformation states are then used to estimate breathing volume through a per-patient correlation metric acquired through a Bayesian-network learning process. Through prototyping and implementation, our results indicate that we have achieved 92.2% to 94.19% accuracy in the tidal volume estimations through the experimentation based on the proposed vision-based method.

  Authors:Michael Mannino, Joel Fredrickson, Farnoush Banaei-Kashani, Iris Linck, Raghda Alqurashi Raghda

  Publication:Transactions on Management Information Systems (TMIS) 2015

  Category: Conference

  Acceptance Rate: 33%

  Link: @ACM

Abstract:We develop a similarity measure for medical event sequences (MESs) and empirically evaluate it using U.S. Medicare claims data. Existing similarity measures do not use unique characteristics of MESs and have never been evaluated on real MESs. Our similarity measure, the Optimal Temporal Common Subsequence for Medical Event Sequences (OTCS-MES), provides a matching component that integrates event prevalence, event duplication, and hierarchical coding, important elements of MESs. The OTCS-MES also uses normalization to mitigate the impact of heavy positive skew of matching events and compact distribution of event prevalence. We empirically evaluate the OTCS-MES measure against two other measures specifically designed for MESs, the original OTCS and Artemis, a measure incorporating event alignment. Our evaluation uses two substantial data sets of Medicare claims data containing inpatient and outpatient sequences with different medical event coding. We find a small overlap in nearest neighbors among the three similarity measures, demonstrating the superior design of the OTCS-MES with its emphasis on unique aspects of MESs. The evaluation also provides evidence about the impact of component weights, neighborhood size, and sequence length.