Research and Publications - Department of Computer Science

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 20 of 26
  • Item
    Editorial: deep learning for 5G IoT systems
    (Springer, 2021) Cheng, Xiaochun; Zhang, Chengqi; Qian, Yi; Aloqaily, Moayad; Xiao, Yang; Middlesex University; University of Technology Sydney; University of Nebraska Lincoln; Qatar University; University of Alabama Tuscaloosa
  • Item
    Food/Non-Food Classification of Real-Life Egocentric Images in Low- and Middle-Income Countries Based on Image Tagging Features
    (Frontiers, 2021) Chen, Guangzong; Jia, Wenyan; Zhao, Yifan; Mao, Zhi-Hong; Lo, Benny; Anderson, Alex K.; Frost, Gary; Jobarteh, Modou L.; McCrory, Megan A.; Sazonov, Edward; Steiner-Asiedu, Matilda; Ansong, Richard S.; Baranowski, Thomas; Burke, Lora; Sun, Mingui; University of Pittsburgh; Imperial College London; University of Georgia; Boston University; University of Alabama Tuscaloosa; University of Ghana; United States Department of Agriculture (USDA); Baylor College of Medicine
    Malnutrition, including both undernutrition and obesity, is a significant problemin low- and middle-income countries (LMICs). In order to study malnutrition and develop effective intervention strategies, it is crucial to evaluate nutritional status in LMICs at the individual, household, and community levels. In a multinational research project supported by the Bill & Melinda Gates Foundation, we have been using a wearable technology to conduct objective dietary assessment in sub-Saharan Africa. Our assessment includes multiple diet-related activities in urban and rural families, including food sources (e.g., shopping, harvesting, and gathering), preservation/storage, preparation, cooking, and consumption (e.g., portion size and nutrition analysis). Our wearable device ("eButton" worn on the chest) acquires real-life images automatically during wake hours at preset time intervals. The recorded images, in amounts of tens of thousands per day, are post-processed to obtain the information of interest. Although we expect future Artificial Intelligence (AI) technology to extract the information automatically, at present we utilize AI to separate the acquired images into two binary classes: images with (Class 1) and without (Class 0) edible items. As a result, researchers need only to study Class-1 images, reducing their workload significantly. In this paper, we present a composite machine learning method to perform this classification, meeting the specific challenges of high complexity and diversity in the real-world LMIC data. Our method consists of a deep neural network (DNN) and a shallow learning network (SLN) connected by a novel probabilistic network interface layer. After presenting the details of our method, an image dataset acquired from Ghana is utilized to train and evaluate the machine learning system. Our comparative experiment indicates that the new composite method performs better than the conventional deep learning method assessed by integrated measures of sensitivity, specificity, and burden index, as indicated by the Receiver Operating Characteristic (ROC) curve.
  • Item
    SoSyM reflections: the 2020 "State of the Journal" report
    (Springer, 2021) Ergin, Huseyin; Gray, Jeff; Rumpe, Bernhard; Schindler, Martin; Ball State University; University of Alabama Tuscaloosa; RWTH Aachen University
  • Item
    SoSyM reflections: the 2021 "state of the journal" report
    (Springer, 2022) Ergin, Huseyin; Gray, Jeff; Rumpe, Bernhard; Schindler, Martin; Ball State University; University of Alabama Tuscaloosa; RWTH Aachen University
  • Item
    Weak Galerkin methods for second order elliptic interface problems
    (Elsevier, 2013) Mu, Lin; Wang, Junping; Wei, Guowei; Ye, Xiu; Zhao, Shan; Michigan State University; National Science Foundation (NSF); NSF - Directorate for Mathematical & Physical Sciences (MPS); NSF - Division of Mathematical Sciences (DMS); University of Arkansas Little Rock; University of Arkansas Fayetteville; University of Alabama Tuscaloosa
    Weak Galerkin methods refer to general finite element methods for partial differential equations (PDEs) in which differential operators are approximated by their weak forms as distributions. Such weak forms give rise to desirable flexibilities in enforcing boundary and interface conditions. A weak Galerkin finite element method (WG-FEM) is developed in this paper for solving elliptic PDEs with discontinuous coefficients and interfaces. Theoretically, it is proved that high order numerical schemes can be designed by using the WG-FEM with polynomials of high order on each element. Extensive numerical experiments have been carried out to validate the WG-FEM for solving second order elliptic interface problems. High order of convergence is numerically confirmed in both L-2 and L-infinity norms for the piecewise linear WG-FEM. Special attention is paid to solve many interface problems, in which the solution possesses a certain singularity due to the nonsmoothness of the interface. A challenge in research is to design nearly second order numerical methods that work well for problems with low regularity in the solution. The best known numerical scheme in the literature is of order O(h) to O(h(1.5)) for the solution itself in L-infinity norm. It is demonstrated that the WG-FEM of the lowest order, i.e., the piecewise constant WG-FEM, is capable of delivering numerical approximations that are of order O(h(1.75)) to O(h(2)) in the L-infinity norm for C-1 or Lipschitz continuous interfaces associated with a C-1 or H-2 continuous solution. (C) 2013 Elsevier Inc. All rights reserved.
  • Item
    Electronic medical records and physician stress in primary care: results from the MEMO Study
    (Oxford University Press, 2014) Babbott, Stewart; Manwell, Linda Baier; Brown, Roger; Montague, Enid; Williams, Eric; Schwartz, Mark; Hess, Erik; Linzer, Mark; University of Kansas; University of Kansas Medical Center; University of Wisconsin Madison; Northwestern University; University of Alabama Tuscaloosa; New York University; Mayo Clinic; Hennepin County Medical Center
    Background Little has been written about physician stress that may be associated with electronic medical records (EMR). Objective We assessed relationships between the number of EMR functions, primary care work conditions, and physician satisfaction, stress and burnout. Design and participants 379 primary care physicians and 92 managers at 92 clinics from New York City and the upper Midwest participating in the 2001-5 Minimizing Error, Maximizing Outcome (MEMO) Study. A latent class analysis identified clusters of physicians within clinics with low, medium and high EMR functions. Main measures We assessed physician-reported stress, burnout, satisfaction, and intent to leave the practice, and predictors including time pressure during visits. We used a two-level regression model to estimate the mean response for each physician cluster to each outcome, adjusting for physician age, sex, specialty, work hours and years using the EMR. Effect sizes (ES) of these relationships were considered small (0.14), moderate (0.39), and large (0.61). Key results Compared to the low EMR cluster, physicians in the moderate EMR cluster reported more stress (ES 0.35, p=0.03) and lower satisfaction (ES -0.45, p=0.006). Physicians in the high EMR cluster indicated lower satisfaction than low EMR cluster physicians (ES -0.39, p=0.01). Time pressure was associated with significantly more burnout, dissatisfaction and intent to leave only within the high EMR cluster. Conclusions Stress may rise for physicians with a moderate number of EMR functions. Time pressure was associated with poor physician outcomes mainly in the high EMR cluster. Work redesign may address these stressors.
  • Item
    How to define modeling languages?
    (Springer, 2023) Combemale, Benoit; Gray, Jeff; Rumpe, Bernhard; Universite de Rennes; University of Alabama Tuscaloosa; RWTH Aachen University
  • Item
    A survey of the state of the practice for research software in the United States
    (PeerJ, 2022) Carver, Jeffrey C.; Weber, Nic; Ram, Karthik; Gesing, Sandra; Katz, Daniel S.; University of Alabama Tuscaloosa; University of Washington; University of Washington Seattle; University of California Berkeley; University of Illinois Urbana-Champaign
    Research software is a critical component of contemporary scholarship. Yet, most research software is developed and managed in ways that are at odds with its long-term sustainability. This paper presents findings from a survey of 1,149 researchers, primarily from the United States, about sustainability challenges they face in developing and using research software. Some of our key findings include a repeated need for more opportunities and time for developers of research software to receive training. These training needs cross the software lifecycle and various types of tools. We also identified the recurring need for better models of funding research software and for providing credit to those who develop the software so they can advance in their careers. The results of this survey will help inform future infrastructure and service support for software developers and users, as well as national research policy aimed at increasing the sustainability of research software.
  • Item
    Automatic Ingestion Monitor Version 2 - A Novel Wearable Device for Automatic Food Intake Detection and Passive Capture of Food Images
    (IEEE, 2021) Doulah, Abul; Ghosh, Tonmoy; Hossain, Delwar; Imtiaz, Masudul H.; Sazonov, Edward; University of Alabama Tuscaloosa
    Use of food image capture and/or wearable sensors for dietary assessment has grown in popularity. Active - methods rely on the user to take an image of each eating episode. "Passive" methods use wearable cameras that continuously capture images. Most of "passively" captured images are not related to food consumption and may present privacy concerns. In this paper, we propose a novel wearable sensor (Automatic Ingestion Monitor. AIM-2) designed to capture images only during automatically detected eating episodes. The capture method was validated on a dataset collected from 30 volunteers in the community wearing the AIM-2 for 24h in pseudo-free-living and 24h in a free-living environment. The AIM-2 was able to detect food intake over 10-second epochs with a (mean and standard deviation) Fl-score of 81.8 +/- 10.1%. The accuracy of eating episode detection was 82.7%. Out of a total of 180,570 images captured, 8,929 (4.9%) images belonged to detected eating episodes. Privacy concerns were assessed by a questionnaire on a scale 1-7. Continuous capture had concern value of 5.0 +/- 1.6 (concerned) while image capture only during food intake had concern value of 1.9 +/- 1.7 (not concerned). Results suggest that AIM-2 can provide accurate detection of food intake, reduce the number of images for analysis and alleviate the privacy concerns of the users.
  • Item
    Automatic Count of Bites and Chews From Videos of Eating Episodes
    (IEEE, 2020) Hossain, Delwar; Ghosh, Tonmoy; Sazonov, Edward; University of Alabama Tuscaloosa
    Methods for measuring of eating behavior (known as meal microstructure) often rely on manual annotation of bites, chews, and swallows on meal videos or wearable sensor signals. The manual annotation may be time consuming and erroneous, while wearable sensors may not capture every aspect of eating (e.g. chews only). The aim of this study is to develop a method to detect and count bites and chews automatically from meal videos. The method was developed on a dataset of 28 volunteers consuming unrestricted meals in the laboratory under video observation. First, the faces in the video (regions of interest, ROI) were detected using Faster R-CNN. Second, a pre-trained AlexNet was trained on the detected faces to classify images as a bite/no bite image. Third, the affine optical flow was applied in consecutively detected faces to find the rotational movement of the pixels in the ROIs. The number of chews in a meal video was counted by converting the 2-D images to a 1-D optical flow parameter and finding peaks. The developed bite and chew count algorithm was applied to 84 meal videos collected from 28 volunteers. A mean accuracy (+/- STD) of 85.4% (+/- 6.3%) with respect to manual annotation was obtained for the number of bites and 88.9% (+/- 7.4%) for the number of chews. The proposed method for an automatic bite and chew counting shows promising results that can be used as an alternative solution to manual annotation.
  • Item
    Selective Content Removal for Egocentric Wearable Camera in Nutritional Studies
    (IEEE, 2020) Abul Hassan, Mohamed; Sazonov, Edward; University of Alabama Tuscaloosa
    Automatic Ingestion Monitor v2 (AIM-2) is an egocentric camera and sensor that aids monitoring of individual diet and eating behavior by capturing still images throughout the day and using sensor data to detect eating. The images may be used to recognize foods being eaten, eating environment, and other behaviors and daily activities. At the same time, captured images may carry privacy concerning content such as (1) people in social eating and/or bystanders (i.e., bystander privacy); (2) sensitive documents that may appear on a computer screen in the view of AIM-2 (i.e., context privacy). In this paper, we propose a novel approach based on automatic, image redaction for privacy protection by selective content removal by semantic segmentation using a deep learning neural network. The proposed method reported a bystander privacy removal with precision of 0.87 and recall of 0.94 and reported context privacy removal by precision and recall of 0.97 and 0.98. The results of the study showed that selective content removal using deep learning neural network is a much more desirable approach to address privacy concerns for an egocentric wearable camera for nutritional studies.
  • Item
    Pneumatic Variable Series Elastic Actuator
    (ASME, 2016) Zheng, Hao; Wu, Molei; Shen, Xiangrong; University of Alabama Tuscaloosa
    Inspired by human motor control theory, stiffness control is highly effective in manipulation and human-interactive tasks. The implementation of stiffness control in robotic systems, however, has largely been limited to closed-loop control, and suffers from multiple issues such as limited frequency range, potential instability, and lack of contribution to energy efficiency. Variable-stiffness actuator represents a better solution, but the current designs are complex, heavy, and bulky. The approach in this paper seeks to address these issues by using pneumatic actuator as a variable series elastic actuator (VSEA), leveraging the compressibility of the working fluid. In this work, a pneumatic actuator is modeled as an elastic element with controllable stiffness and equilibrium point, both of which are functions of air masses in the two chambers. As such, for the implementation of stiffness control in a robotic system, the desired stiffness/equilibrium point can be converted to the desired chamber air masses, and a predictive pressure control approach is developed to control the timing of valve switching to obtain the desired air mass while minimizing control action. Experimental results showed that the new approach in this paper requires less expensive hardware (on-off valve instead of proportional valve), causes less control action in implementation, and provides good control performance by leveraging the inherent dynamics of the actuator.
  • Item
    Individuals' privacy concerns and adoption of contact tracing mobile applications in a pandemic: A situational privacy calculus perspective
    (Oxford University Press, 2021) Hassandoust, Farkhondeh; Akhlaghpour, Saeed; Johnston, Allen C.; Auckland University of Technology; University of Queensland; University of Alabama Tuscaloosa
    Objective: The study sought to develop and empirically validate an integrative situational privacy calculus model for explaining potential users' privacy concerns and intention to install a contact tracing mobile application (CTMA). Materials and Methods: A survey instrument was developed based on the extant literature in 2 research streams of technology adoption and privacy calculus. Survey participants (N = 853) were recruited from all 50 U.S. states. Partial least squares structural equation modeling was used to validate and test the model. Results: Individuals' intention to install a CTMA is influenced by their risk beliefs, perceived individual and societal benefits to public health, privacy concerns, privacy protection initiatives (legal and technical protection), and technology features (anonymity and use of less sensitive data). We found only indirect relationships between trust in public health authorities and intention to install CTMA. Sex, education, media exposure, and past invasion of privacy did not have a significant relationship either, but interestingly, older people were slightly more inclined than younger people to install a CTMA. Discussion: Our survey results confirm the initial concerns about the potentially low adoption rates of CTMA. Our model provides public health agencies with a validated list of factors influencing individuals' privacy concerns and beliefs, enabling them to systematically take actions to address these identified issues, and increase CTMA adoption. Conclusions: Developing CTMAs and increasing their adoption is an ongoing challenge for public health systems and policymakers. This research provides an evidence-based and situation-specific model for a better understanding of this theoretically and pragmatically important phenomenon.
  • Item
    Segmentation and Characterization of Chewing Bouts by Monitoring Temporalis Muscle Using Smart Glasses With Piezoelectric Sensor
    (IEEE, 2017) Farooq, Muhammad; Sazonov, Edward; University of Alabama Tuscaloosa
    Several methods have been proposed for automatic and objective monitoring of food intake, but their performance suffers in the presence of speech and motion artifacts. This paper presents a novel sensor system and algorithms for detection and characterization of chewing bouts from a piezoelectric strain sensor placed on the temporalis muscle. The proposed data acquisition device was incorporated into the temple of eyeglasses. The system was tested by ten participants in two part experiments, one under controlled laboratory conditions and the other in unrestricted free-living. The proposed food intake recognition method first performed an energy-based segmentation to isolate candidate chewing segments (instead of using epochs of fixed duration commonly reported in research literature), with the subsequent classification of the segments by linear support vector machine models. On participant level (combining data from both laboratory and free-living experiments), with ten-fold leave-one-out cross-validation, chewing were recognized with average F-score of 96.28% and the resultant area under the curve was 0.97, which are higher than any of the previously reported results. A multivariate regression model was used to estimate chew counts from segments classified as chewing with an average mean absolute error of 3.83% on participant level. These results suggest that the proposed system is able to identify chewing segments in the presence of speech and motion artifacts, as well as automatically and accurately quantify chewing behavior, both under controlled laboratory conditions and unrestricted free-living.
  • Item
    Posture and Activity Recognition and Energy Expenditure Estimation in a Wearable Platform
    (IEEE, 2015) Sazonov, Edward; Hegde, Nagaraj; Browning, Raymond C.; Melanson, Edward L.; Sazonova, Nadezhda A.; University of Alabama Tuscaloosa; Colorado State University; University of Colorado Denver
    The use of wearable sensors coupled with the processing power of mobile phones may be an attractive way to provide real-time feedback about physical activity and energy expenditure (EE). Here, we describe the use of a shoe-based wearable sensor system (SmartShoe) with a mobile phone for real-time recognition of various postures/physical activities and the resulting EE. To deal with processing power and memory limitations of the phone, we compare the use of support vector machines (SVM), multinomial logistic discrimination (MLD), and multilayer perceptrons (MLP) for posture and activity classification followed by activity-branched EE estimation. The algorithms were validated using data from 15 subjects who performed up to 15 different activities of daily living during a 4-h stay in a room calorimeter. MLD and MLP demonstrated activity classification accuracy virtually identical to SVM (similar to 95%) while reducing the running time and the memory requirements by a factor of >10(3). Comparison of per-minute EE estimation using activity-branched models resulted in accurate EE prediction (RMSE = 0.78 kcal/min for SVM andMLD activity classification, 0.77 kcal/min for MLP versus RMSE of 0.75 kcal/min for manual annotation). These results suggest that low-power computational algorithms can be successfully used for real-time physical activity monitoring and EE estimation on a wearable platform.
  • Item
    Bayesian analysis of the impact of rainfall data product on simulated slope failure for North Carolina locations
    (Springer, 2019) Yatheendradas, Soni; Kirschbaum, Dalia; Nearing, Grey; Vrugt, Jasper A.; Baum, Rex L.; Wooten, Rick; Lu, Ning; Godt, Jonathan W.; University of Maryland College Park; National Aeronautics & Space Administration (NASA); NASA Goddard Space Flight Center; University of Alabama Tuscaloosa; University of California Irvine; United States Department of the Interior; United States Geological Survey
    In the past decades, many different approaches have been developed in the literature to quantify the load-carrying capacity and geotechnical stability (or the factor of safety, F-s) of variably saturated hillslopes. Much of this work has focused on a deterministic characterization of hillslope stability. Yet, simulated F-s values are subject to considerable uncertainty due to our inability to characterize accurately the soil mantle's properties (hydraulic, geotechnical, and geomorphologic) and spatiotemporal variability of the moisture content of the hillslope interior. This is particularly true at larger spatial scales. Thus, uncertainty-incorporating analyses of physically based models of rain-induced landslides are rare in the literature. Such landslide modeling is typically conducted at the hillslope scale using gauge-based rainfall forcing data with rather poor spatiotemporal coverage. For regional landslide modeling, the specific advantages and/or disadvantages of gauge-only, radar-merged and satellite-based rainfall products are not clearly established. Here, we compare and evaluate the performance of the Transient Rainfall Infiltration and Grid-based Regional Slope-stability analysis (TRIGRS) model for three different rainfall products using 112 observed landslides in the period between 2004 and 2011 from the North Carolina Geological Survey database. Our study includes the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis Version 7 (TMPA V7), the North American Land Data Assimilation System Phase 2 (NLDAS-2) analysis, and the reference truth Stage IV precipitation. TRIGRS model performance was rather inferior with the use of literature values of the geotechnical parameters and soil hydraulic properties from ROSETTA using soil textural and bulk density data from SSURGO (Soil Survey Geographic database). The performance of TRIGRS improved considerably after Bayesian estimation of the parameters with the DiffeRential Evolution Adaptive Metropolis (DREAM) algorithm using Stage IV precipitation data. Hereto, we use a likelihood function that combines binary slope failure information from landslide event and null periods using multivariate frequency distribution-based metrics such as the false discovery and false omission rates. Our results demonstrate that the Stage IV-inferred TRIGRS parameter distributions generalize well to TMPA and NLDAS-2 precipitation data, particularly at sites with considerably larger TMPA and NLDAS-2 rainfall amounts during landslide events than null periods. TRIGRS model performance is then rather similar for all three rainfall products. At higher elevations, however, the TMPA and NLDAS-2 precipitation volumes are insufficient and their performance with the Stage IV-derived parameter distributions indicates their inability to accurately characterize hillslope stability.
  • Item
    Understanding the experiences of self-injurious behavior in autism spectrum disorder: Implications for monitoring technology design
    (Oxford University Press, 2021) Cantin-Garside, Kristine D.; Nussbaum, Maury A.; White, Susan W.; Kim, Sunwook; Kim, Chung Do; Fortes, Diogo M. G.; Valdez, Rupa S.; Virginia Polytechnic Institute & State University; University of Virginia; University of Alabama Tuscaloosa
    Objective: Monitoring technology may assist in managing self-injurious behavior (SIB), a pervasive concern in autism spectrum disorder (ASD). Affiliated stakeholder perspectives should be considered to design effective and accepted SIB monitoring methods. We examined caregiver experiences to generate design guidance for SIB monitoring technology. Materials and Methods: Twenty-three educators and 16 parents of individuals with ASD and SIB completed interviews or focus groups to discuss needs related to monitoring SIB and associated technology use. Results: Qualitative content analysis of participant responses revealed 7 main themes associated with SIB and technology: triggers, emotional responses, SIB characteristics, management approaches, caregiver impact, child/student impact, and sensory/technology preferences. Discussion: The derived themes indicated areas of emphasis for design at the intersection of monitoring and SIB. Systems design at this intersection should consider the range of manifestations of and management approaches for SIB. It should also attend to interactions among children with SIB, their caregivers, and the technology. Design should prioritize the transferability of physical technology and behavioral data as well as the safety, durability, and sensory implications of technology. Conclusions: The collected stakeholder perspectives provide preliminary groundwork for an SIB monitoring system responsive to needs as articulated by caregivers. Technology design based on this groundwork should follow an iterative process that meaningfully engages caregivers and individuals with SIB in naturalistic settings.
  • Item
    A Systematic Review of Technology-Driven Methodologies for Estimation of Energy Intake
    (IEEE, 2019) Doulah, Abul; Mccrory, Megan A.; Higgins, Janine A.; Sazonov, Edward; University of Alabama Tuscaloosa; Boston University; University of Colorado Denver
    Accurate measurement of energy intake (EI) is important for estimation of energy balance, and, correspondingly, body weight dynamics. Traditional measurements of EI rely on self-report, which may be inaccurate and underestimate EI. The imperfections in traditional methodologies such as 24-hour dietary recall, dietary record, and food frequency questionnaire stipulate development of technology-driven methods that rely on wearable sensors and imaging devices to achieve an objective and accurate assessment of EI. The aim of this research was to systematically review and examine peer-reviewed papers that cover the estimation of EI in humans, with the focus on emerging technology-driven methodologies. Five major electronic databases were searched for articles published from January 2005 to August 2017: Pubmed, Science Direct, IEEE Xplore, ACM library, and Google Scholar. Twenty-six eligible studies were retrieved that met the inclusion criteria. The review identified that while the current methods of estimating EI show promise, accurate estimation of EI in free-living individuals presents many challenges and opportunities. The most accurate result identified for EI (kcal) estimation had an average accuracy of 94%. However, collectively, the results were obtained from a limited number of food items (i.e., 19), small sample sizes (i.e., 45 meal images), and primarily controlled conditions. Therefore, new methods that accurately estimate EI over long time periods in free-living conditions are needed.
  • Item
    A Machine-Learning Based Approach for Predicting Older Adults? Adherence to Technology-Based Cognitive Training
    (Elsevier, 2022) He, Zhe; Tian, Shubo; Singh, Ankita; Chakraborty, Shayok; Zhang, Shenghao; Lustria, Mia Liza A.; Charness, Neil; Roque, Nelson A.; Harrell, Erin R.; Boot, Walter R.; Florida State University; University of Central Florida; University of Alabama Tuscaloosa
    Adequate adherence is a necessary condition for success with any intervention, including for computerized cognitive training designed to mitigate age-related cognitive decline. Tailored prompting systems offer promise for promoting adherence and facilitating intervention success. However, developing adherence support systems capable of just-in-time adaptive reminders re-quires understanding the factors that predict adherence, particularly an imminent adherence lapse. In this study we built machine learning models to predict participants' adherence at different levels (overall and weekly) using data collected from a previous cognitive training intervention. We then built machine learning models to predict adherence using a variety of baseline measures (demographic, attitudinal, and cognitive ability variables), as well as deep learning models to predict the next week's adherence using variables derived from training in-teractions in the previous week. Logistic regression models with selected baseline variables were able to predict overall adherence with moderate accuracy (AUROC: 0.71), while some recurrent neural network models were able to predict weekly adherence with high accuracy (AUROC: 0.84-0.86) based on daily interactions. Analysis of the post hoc explanation of machine learning models revealed that general self-efficacy, objective memory measures, and technology self-efficacy were most predictive of participants' overall adherence, while time of training, ses-sions played, and game outcomes were predictive of the next week's adherence. Machine-learning based approaches revealed that both individual difference characteristics and previous inter-vention interactions provide useful information for predicting adherence, and these insights can provide initial clues as to who to target with adherence support strategies and when to provide support. This information will inform the development of a technology-based, just-in-time adherence support systems.
  • Item
    Wineinformatics: Using the Full Power of the Computational Wine Wheel to Understand 21st Century Bordeaux Wines from the Reviews
    (MDPI, 2021) Dong, Zeqing; Atkison, Travis; Chen, Bernard; University of Central Arkansas; University of Alabama Tuscaloosa
    Although wine has been produced for several thousands of years, the ancient beverage has remained popular and even more affordable in modern times. Among all wine making regions, Bordeaux, France is probably one of the most prestigious wine areas in history. Since hundreds of wines are produced from Bordeaux each year, humans are not likely to be able to examine all wines across multiple vintages to define the characteristics of outstanding 21st century Bordeaux wines. Wineinformatics is a newly proposed data science research with an application domain in wine to process a large amount of wine data through the computer. The goal of this paper is to build a high-quality computational model on wine reviews processed by the full power of the Computational Wine Wheel to understand 21st century Bordeaux wines. On top of 985 binary-attributes generated from the Computational Wine Wheel in our previous research, we try to add additional attributes by utilizing a CATEGORY and SUBCATEGORY for an additional 14 and 34 continuous-attributes to be included in the All Bordeaux (14,349 wine) and the 1855 Bordeaux datasets (1359 wines). We believe successfully merging the original binary-attributes and the new continuous-attributes can provide more insights for Naive Bayes and Supported Vector Machine (SVM) to build the model for a wine grade category prediction. The experimental results suggest that, for the All Bordeaux dataset, with the additional 14 attributes retrieved from CATEGORY, the Naive Bayes classification algorithm was able to outperform the existing research results by increasing accuracy by 2.15%, precision by 8.72%, and the F-score by 1.48%. For the 1855 Bordeaux dataset, with the additional attributes retrieved from the CATEGORY and SUBCATEGORY, the SVM classification algorithm was able to outperform the existing research results by increasing accuracy by 5%, precision by 2.85%, recall by 5.56%, and the F-score by 4.07%. The improvements demonstrated in the research show that attributes retrieved from the CATEGORY and SUBCATEGORY has the power to provide more information to classifiers for superior model generation. The model build in this research can better distinguish outstanding and class 21st century Bordeaux wines. This paper provides new directions in Wineinformatics for technical research in data science, such as regression, multi-target, classification and domain specific research, including wine region terroir analysis, wine quality prediction, and weather impact examination.