ROLEX: A Novel Method for Interpretable Machine Learning Using Robust Local Explanations

2023 | Management Information Systems Quarterly | Citations: 2

Authors: Kim, Buomsoo (Raymond); Srinivasan, Karthik; Kong, Sung Hye; Kim, Jung Hee; Shin, Chan Soo; Ram, Sudha

Abstract: Recent developments in big data technologies are revolutionizing the field of h ... Expand

Abstract: Recent developments in big data technologies are revolutionizing the field of healthcare predictive analytics (HPA), enabling researchers to explore challenging problems using complex prediction models. Nevertheless, healthcare practitioners are reluctant to adopt those models as they are less transparent and accountable due to their black-box structure. We believe that instance-level, or local, explanations enhance patient safety and foster trust by enabling patient-level interpretations and medical knowledge discovery. Therefore, we propose the RObust Local EXplanations (ROLEX) method to develop robust, instance-level explanations for HPA models in this study. ROLEX adapts state-of-the-art methods and ameliorates their shortcomings in explaining individual-level predictions made by black-box machine learning models. Our analysis with a large real-world dataset related to a prevalent medical condition called fragility fracture and two publicly available healthcare datasets reveals that ROLEX outperforms widely accepted benchmark methods in terms of local faithfulness of explanations. In addition, ROLEX is more robust since it does not rely on extensive hyperparameter tuning or heuristic algorithms. Explanations generated by ROLEX, along with the prototype user interface presented in this study, have the potential to promote personalized care and precision medicine by providing patient-level interpretations and novel insights. We discuss the theoretical implications of our study in healthcare, big data, and design science. Collapse

Topics: explainable artificial intelligence usability user interface local interpretable model agnostic explanation problem solving

Methods: machine learning design artifact computational algorithm longitudinal research ablation analysis

Theories: lemon market theory

The challenges of entering the metaverse: An experiment on the effect of extended reality on workload

2023 | Information Systems Frontiers | Citations: 4

Authors: Xi, Nannan; Chen, Juan; Gama, Filipe; Riar, Marc; Hamari, Juho

Abstract: Information technologies exist to enable us to either do things we have not done ... Expand

Abstract: Information technologies exist to enable us to either do things we have not done before or do familiar things more efficiently. Metaverse (i.e. extended reality: XR) enables novel forms of engrossing telepresence, but it also may make mundate tasks more effortless. Such technologies increasingly facilitate our work, education, healthcare, consumption and entertainment; however, at the same time, metaverse bring a host of challenges. Therefore, we pose the question whether XR technologies, specifically Augmented Reality (AR) and Virtual Reality (VR), either increase or decrease the difficulties of carrying out everyday tasks. In the current study we conducted a 2 (AR: with vs. without) × 2 (VR: with vs. without) between-subject experiment where participants faced a shopping-related task (including navigating, movement, hand-interaction, information processing, information searching, storing, decision making, and simple calculation) to examine a proposed series of hypotheses. The NASA Task Load Index (NASA-TLX) was used to measure subjective workload when using an XR-mediated information system including six sub-dimensions of frustration, performance, effort, physical, mental, and temporal demand. The findings indicate that AR was significantly associated with overall workload, especially mental demand and effort, while VR had no significant effect on any workload sub-dimensions. There was a significant interaction effect between AR and VR on physical demand, effort, and overall workload. The results imply that the resources and cost of operating XR-mediated realities are different and higher than physical reality. Collapse

Topics: virtual reality extended reality usability ease of use augmented reality

Methods: experiment parametric test multivariate analysis of variance ANOVA survey

Theories: information systems theory

Making Filter Bubbles Understandable

2022 | Americas Conference on Information Systems | Citations: 0

Authors: Hirschmeier, Stefan; Ockenga, Tim Alvaro

Abstract: Recommender systems tend to create filter bubbles and, as a consequence, lower ... Expand

Abstract: Recommender systems tend to create filter bubbles and, as a consequence, lower diversity exposure, often with the user not being aware of it. The biased preselection of content by recommender systems has called for approaches to deal with exposure diversity, such as giving users control over their filter bubble. We analyze how to make filter bubbles understandable and controllable by using interactive word clouds, following the idea of building trust in the system. On the basis of several prototypes, we performed explorative research on how to design word clouds for the controllability of filter bubbles. Our findings can inform designers of interactive filter bubbles in personalized offers of broadcasters, publishers, and media houses. Collapse

Topics: cloud computing filter bubble recommender system smartphone interactive interface

Methods: design artifact qualitative interview cognitive walkthrough design science survey

How does IT affect design centricity approaches: Evidence from Spain’s smart tourism ecosystem

2019 | International Journal of Information Management | Citations: 0

Authors: Arenas, Alvaro E.; Goh, Jie Mein; Urueña, Alberto

Abstract: Little or no prior work has examined how information technology enables the deve ... Expand

Abstract: Little or no prior work has examined how information technology enables the development of a design centered digital ecosystem. To examine this research question, we employ a capabilities lens and identify the pathways through which IT drives the development of a design centric smart tourism ecosystem. We analyzed archival data and data collected from interviews conducted in Spain, a country which has embarked on smart destinations projects and topped the World Economic Forum’s Travel and Competitiveness Index. From our analysis, we delineate and identify specific IT-enabled capabilities important for a country implementing smart tourism projects. We find that many of the IT resources available help develop key capabilities necessary for creating a design centric smart tourism ecosystem. Collapse

Topics: information technology capability social capital interactive interface human centered design enabling capability

Methods: qualitative interview design methodology archival research personal interview theoretical contribution

Human Interpretation of Trade-Off Diagrams in Multi-Objective Problems: Implications for Developing Interactive Decision Support Systems

2019 | HICSS | Citations: 0

Authors: Oprean, Danielle; Spence, Caitlin; Simpson, Mark; Miller, Randy; Bansal, Saurabh; Keller, Klaus; Klippel, Alexander

Abstract: The growing need for efficient and effective human decision-makers warrants a be ... Expand

Abstract: The growing need for efficient and effective human decision-makers warrants a better understanding of how decision support systems (DSS) guide users to improved decisions. Decision support approaches utilize visual aids to assist decision-making, including trade-off diagrams. These visualizations help comprehension of key trade-offs among decision alternatives. However, little is known about the role of trade-off diagrams in human decision-making and the best way to present them. Here, we discuss an empirical study with two goals: 1) evaluating DSS interactivity and 2) identifying decision-making strategies with trade-off diagrams. We specifically investigate the value of interface interactivity and problem context as users make nine increasingly complex decisions. Our results suggest that problem context and interactivity separately influence ability to navigate trade-off diagrams. Collapse

Topics: decision support system decision making decision quality interactive interface usability

Methods: system dynamics experiment multivariate analysis of variance two-way ANOVA parametric test

Theories: theory of affordance

How younger elderly realize usefulness of cognitive training video games to maintain their independent living

2018 | International Journal of Information Management | Citations: 0

Authors: Talaei-Khoei, Amir; Daniel, Jay

Abstract: The objective of this paper is to understand the perception that younger elderly ... Expand

Abstract: The objective of this paper is to understand the perception that younger elderly persons have towards the usefulness of playing Xbox Kinect video games as an assistive technology that is designed to maintain their cognitive abilities. Available literature highlights two kinds of assistive technologies; the first being Supportive Technologies that provide aid for already-declined functional abilities (such as hearing aids), and the second being Empowering Technologies that maintain functional abilities which have not yet declined (such as Xbox Kinect cognitive games). The difference in the nature between supportive and empowering technologies plays an important role in perceiving their benefits. For instance, while hearing aids as a supportive technology are perceived as useful through the improvement of hearing abilities, cognitive training games as an empowering technology have a long-term usefulness for cognitive abilities. This study conducts twenty-one qualitative interviews (range 65–87 years; mean = 71; SD = 3.81) and introduces perceived transfer effect. This effect allows the elderly to perceive the usefulness of playing cognitive training video games, which are designed to cultivate the cognitive abilities. In addition, this study found that the elderly value their independent living, and through cognitive video games, the elderly may remain capable of living independently. Collapse

Topics: video game robotic perceived usefulness interactive interface information system use

Methods: qualitative interview design artifact personal interview conceptual modelling theoretical contribution

Theories: technology acceptance theory

Data Visualization on Global Trends on Cancer Incidence An Application of IBM Watson Analytics

2017 | HICSS | Citations: 0

Authors: Tsoi, Kelvin KF; Chan, Felix CH; Hirai, Hoyee W; Leung, Gary KS; Kuo, Yong-Hong; Tai, Samson; Meng, Helen

Abstract: Visual analytics is widely used to explore data patterns and trends. This work l ... Expand

Abstract: Visual analytics is widely used to explore data patterns and trends. This work leverages cancer data collected by World Health Organization (WHO) across over a hundred of cancer registries worldwide. In this study, we present a visual analytics platform, IBM Watson Analytics, to explore the patterns of global cancer incidence. We included 26 cancers from different geographic regions. An interactive interface was applied to plot a choropleth map to show global cancer distribution, and line charts to demonstrate historical cancer trends over 29 years. Subgroup analyses were conducted for different age groups. With real-time interactive features, we can easily explore the data with a selection of any cancer type, gender, age group, or geographical region. This platform is running on the cloud, so it can handle data in huge volumes, and is assessable by any computer connected to the Internet. Collapse

Topics: visual analytics cloud computing data visualization database system digital dashboard

Methods: exploratory data analysis longitudinal research literature study

Space-Time Kernel Density Estimation for Real-Time Interactive Visual Analytics

2017 | HICSS | Citations: 0

Authors: Eaglin, Todd; Cho, Isaac; Ribarsky, William

Abstract: We present a GPU-based implementation of the Space-Time Kernel Density Estimatio ... Expand

Abstract: We present a GPU-based implementation of the Space-Time Kernel Density Estimation (STKDE) that provides massive speed up in analyzing spatialtemporal data. In our work we are able to achieve subsecond performance for data sizes transferable over the Internet in realistic time. We have integrated this into web-based visual interactive analytics tools for analyzing spatial-temporal data. The resulting integrated visual analytics (VA) system permits new analyses of spatial-temporal data from a variety of sources. Novel, interlinked interface elements permit efficient, meaningful analyses. Collapse

Topics: Python hurricane web service internet technology programming language

Methods: computational algorithm usability testing case study simulation survey design

Special issue on leveraging the IS organization for business value creation

2014 | Journal of Information Technology | Citations: 3

Authors: Lee, Gwanhoo; DeLone, William H.; Tan, Margaret; Corrales, Martha

Abstract: ... Expand

Abstract: Collapse

Topics: IT governance value creation IS value IS management government system

Methods: qualitative interview social network analysis case study survey literature sample

Theories: management theory

Dynamic Personal Feedback in Acquiring Information to Manage Your Health

2011 | International Conference on Information Systems | Citations: 0

Authors: Ronen, Hadar; Te'eni, Dov

Abstract: This paper makes two contributions at two different levels: one is about a desig ... Expand

Abstract: This paper makes two contributions at two different levels: one is about a design principle of HCI, namely the provision of personalized and dynamic feedback in interactive applications; the second contribution is a demonstration of the need for visual and dynamic representations to explain the design of interactive interfaces. Collapse

Topics: usability self efficacy human computer interaction personalization data visualization

Methods: survey design principle exploratory data analysis survey design correlation analysis