Eun Kyoung Choe

Eun Kyoung Choe

Auto-generated profile, based on publicly available information

Eun Kyoung Choe

University of Maryland, College Park, Maryland, USA

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 May 2015 in IEEE Computer Graphics and Applications
Verified
Journal Article
Closed access
ECEun Kyoung Choe
Eun Kyoung Choe
BLBongshin Lee
Bongshin Lee
MSM.c. Schraefel
M.c. Schraefel
Data visualization and analytics research has great potential to empower people to improve their lives by leveraging their own personal data. However, most quantified selfers (Q-Selfers) are neither visualization experts nor data scientists. Consequently, visualizations Q-Selfers created with their data are often not ideal for conveying insights. Aiming to design a visualization system to help nonexperts gain and communicate personal data insights, the authors conducted a predesign empirical study. Through the lens of Q-Selfers, they examined what insights people gain specifically from their personal data and how they use visualizations to communicate their insights. Based on their analysis of 30 quantified self-presentations, they characterized eight insight types (detail, self-reflection, trend, comparison, correlation, data summary, distribution, and outlier) and mapped the visual annotations used to communicate them. They further discussed four areas for the design of personal visualization systems, including support for encouraging self-reflection, gaining valid insight, communicating insight, and using visual annotations.

19 August 2019 in IEEE Transactions on Visualization and Computer Graphics
Verified
Journal Article
Closed access
MBMatthew Brehmer
Matthew Brehmer
BLBongshin Lee
Bongshin Lee
PIPetra Isenberg
Petra Isenberg
ECEun Kyoung Choe
Eun Kyoung Choe
We compare the efficacy of animated and small multiples variants of scatterplots on mobile phones for comparing trends in multivariate datasets. Visualization is increasingly prevalent in mobile applications and mobile-first websites, yet there is little prior visualization research dedicated to small displays. In this paper, we build upon previous experimental research carried out on larger displays that assessed animated and non-animated variants of scatterplots. Incorporating similar experimental stimuli and tasks, we conducted an experiment where 96 crowdworker participants performed nine trend comparison tasks using their mobile phones. We found that those using a small multiples design consistently completed tasks in less time, albeit with slightly less confidence than those using an animated design. The accuracy results were more task-dependent, and we further interpret our results according to the characteristics of the individual tasks, with a specific focus on the trajectories of target and distractor data items in each task. We identify cases that appear to favor either animation or small multiples, providing new questions for further experimental research and implications for visualization design on mobile devices. Lastly, we provide a reflection on our evaluation methodology.

20 August 2018 in IEEE Transactions on Visualization and Computer Graphics
Verified
Journal Article
Closed access
MBMatthew Brehmer
Matthew Brehmer
BLBongshin Lee
Bongshin Lee
PIPetra Isenberg
Petra Isenberg
ECEun Kyoung Choe
Eun Kyoung Choe
In the first crowdsourced visualization experiment conducted exclusively on mobile phones, we compare approaches to visualizing ranges over time on small displays. People routinely consume such data via a mobile phone, from temperatures in weather forecasting apps to sleep and blood pressure readings in personal health apps. However, we lack guidance on how to effectively visualize ranges on small displays in the context of different value retrieval and comparison tasks, or with respect to different data characteristics such as periodicity, seasonality, or the cardinality of ranges. Central to our experiment is a comparison between two ways to lay out ranges: a more conventional linear layout strikes a balance between quantitative and chronological scale resolution, while a less conventional radial layout emphasizes the cyclicality of time and may prioritize discrimination between values at its periphery. With results from 87 crowd workers, we found that while participants completed tasks more quickly with linear layouts than with radial ones, there were few differences in terms of error rate between layout conditions. We also found that participants performed similarly with both layouts in tasks that involved comparing superimposed observed and average ranges.

22 December 2020 in Journal of Medical Internet Research
Verified
Journal Article
Open access
YLYuhan Luo
Yuhan Luo
COChi Young Oh
Chi Young Oh
BJBeth St Jean
Beth St Jean
ECEun Kyoung Choe
Eun Kyoung Choe
Background: Although the use of patient-generated data (PGD) in the optimization of patient care shows great promise, little is known about whether patients who track their PGD necessarily share the data with their clinicians. Meanwhile, health literacy—an important construct that captures an individual’s ability to manage their health and to engage with their health care providers—has often been neglected in prior studies focused on PGD tracking and sharing. To leverage the full potential of PGD, it is necessary to bridge the gap between patients’ data tracking and data sharing practices by first understanding the interrelationships between these practices and the factors contributing to these practices. Objective: This study aims to systematically examine the interrelationships between PGD tracking practices, data sharing practices, and health literacy among individual patients. Methods: We surveyed 109 patients at the time they met with a clinician at a university health center, unlike prior research that often examined patients’ retrospective experience after some time had passed since their clinic visit. The survey consisted of 39 questions asking patients about their PGD tracking and sharing practices based on their current clinical encounter. The survey also contained questions related to the participants’ health literacy. All the participants completed the survey on a tablet device. The onsite survey study enabled us to collect ecologically valid data based on patients’ immediate experiences situated within their clinic visit. Results: We found no evidence that tracking PGD was related to self-reports of having sufficient information to manage one’s health; however, the number of data types participants tracked positively related to their self-assessed ability to actively engage with health care providers. Participants’ data tracking practices and their health literacy did not relate to their data sharing practices; however, their ability to engage with health care providers positively related to their willingness to share their data with clinicians in the future. Participants reported several benefits of, and barriers to, sharing their PGD with clinicians. Conclusions: Although tracking PGD could help patients better engage with health care providers, it may not provide patients with sufficient information to manage their health. The gaps between tracking and sharing PGD with health care providers call for efforts to inform patients of how their data relate to their health and to facilitate efficient clinician-patient communication. To realize the full potential of PGD and to promote individuals’ health literacy, empowering patients to effectively track and share their PGD is important—both technologies and health care providers can play important roles.

12 March 2019 in Journal of the American Medical Informatics Association
Verified
Journal Article
Open access
PDF file
YKYoojung Kim
Yoojung Kim
BLBongshin Lee
Bongshin Lee
ECEun Kyoung Choe
Eun Kyoung Choe
Despite the potential values self-tracking data could offer, we have little understanding of how much access people have to “their” data. Our goal of this article is to unveil the current state of the data accessibility—the degree to which people can access their data—of personal health apps in the market. We reviewed 240 personal health apps from the App Store and selected 45 apps that support semi-automated tracking. We characterized the data accessibility of these apps using two dimensions—data access methods and data types. More than 90% of our sample apps (n = 41) provide some types of data access support, which include synchronizing data with a health platform (ie, Apple Health), file download, and application program interfaces. However, the two approachable data access methods for laypeople—health platform and file download—typically put a significant limit on data format, granularity, and amount, which constrains people from easily repurposing the data. Personal data should be accessible to the people who collect them, but existing methods lack sufficient support for people in accessing the fine-grained data. Lack of standards in personal health data schema as well as frequent changes in market conditions are additional hurdles to data accessibility. Many stakeholders including patients, healthcare providers, researchers, third-party developers, and the general public rely on data accessibility to utilize personal data for various goals. As such, improving data accessibility should be considered as an important factor in designing personal health apps and health platforms.

17 May 2019 in JMIR Human Factors
Verified
Journal Article
Open access
ECEun Kyoung Choe
Eun Kyoung Choe
MDMarisa E Duarte
Marisa E Duarte
HSHyewon Suh
Hyewon Suh
WPWanda Pratt
Wanda Pratt
JKJulie A Kientz
Julie A Kientz
JMIR Human Factors - a leading peer-reviewed ehealth journal