Current Research
-
Designing AI Assistant Tool for Dementia Caregivers
This study explores the needs and design insights of personal AI assistant to support dementia caregivers. We interview family caregivers to understand how technologies could be designed to facilitate them taking care of their loves one.
Research Methods: Interview, Qualitative Analysis
Past Research
-
Designing Conversational AI for Older adults
Recent advances in conversational AI and the ubiquity of related devices and applications---from robots to smart speakers to chatbots---has led to extensive research on designing and studying conversational systems with older adults. Despite a growing literature on this topic, many studies examine small groups of older adults and specific devices, neglecting a holistic understanding of how diverse groups of older adults perceive conversational interaction more broadly. We present a systematic review that analyzes older adults' perceptions of and preferences for AI-based conversational systems based on research published from 2010 to mid 2024. Our findings synthesize older adults’ perceptions of the challenges and opportunities for interacting with these systems. We highlight their vision for future AI-based conversational systems, emphasizing a desire for more human-like interactions, personalization, and greater control over their information. We discuss the implications for future research and design of conversational AI systems for older adults.
Research Methods: Systematic Review, PRISMA, Qualitative Analysis
Under review at CHI 2025 -
Co-design self-monitoring technology with healthcare provider
Patient-generated data (PGD) is valuable to providers but is often overwhelmingly large and difficult to use in practice. Researchers have a limited understanding of how tools can surface data insights useful for clinical decision-making. A particular gap is how data-centric technologies can support clinical decision-making for patients with comorbidities. We interviewed 11 providers from Federally Qualified Health Centers (FQHCs) treating patients with metabolic syndrome. Providers wanted patients to collect different data depending on the sets of comorbidities they experienced and patient behaviors. Providers anticipated data values based on social determinants of health. Providers discussed how they would contextualize data events based on medical history and context. We discuss how customized data selection tools, and contextualizing data events with provider heuristics related to patient characteristics can support clinical decision-making..
Research Methods: Co-design, Interview, Qualitative Analysis
Under review at CHI 2025 -
Designing tool for critical reflection on mood
This study seeks to understand if digital tools can promote critical reflection . Critical reflection is classified as transformative reflection which is situated, and cyclic, using multiple perspectives and considering aspects beyond the current context [1, 2]. We conducted a field deployment of two different interactive prototypes that ask people to track their moods .
Research Methods: Survey (closed- and open-ended), Interview, Qualitative Analysis
[1] D. Eisenberg, J. Hunt, N. Speer, and K. Zivin, “Mental Health Service Utilization Among College Students in the United States,”The Journal of Nervous and Mental Disease, vol. 199, no. 5, pp. 301–308, May 2011, doi10.1097/NMD.0b013e3182175123.
[2] R. Fleck and G. Fitzpatrick, “Reflecting on reflection: Framing a design landscape,” Jan. 2010, pp. 216–223. doi: 10.1145/1952222.1952269. -
Current Landscape of Self-Reflection for Mental Health Interventions
In recent years, we have seen a growing prevalence of mental health concerns, resulting in a need for more research on scalable, effective interventions. Self-reflection is an important, evidence-based skill that can improve mental health. In response to the growing demand for digital interventions that support self-reflection and mental health, we review how self-reflection has been conceptualized in HCI research thus far. Using a scoping review, we look at prior work designing and evaluating self-reflection for mental health (SRMH) interventions. We found no uniform definition of “self-reflection,” so we present a five-component definition that allows for a flexible definition. This five- component definition gives researchers and designers the ability to communicate their ideas using a systematic approach, compare components of SRMH interventions, and uncover novel interventions. We present recommendations and avenues of research that will standardized and promote innovation within SRMH interventions and digital mental health tools.
Research Methods: Systematic Review, Grounded Theory, Qualitative AnalysisThis paper is currently under review.
-
Predict the stress level of non-responding day during pregnancy
In this study, we analyzed EMA and physiological data collected from 100 pregnant women over 12 weeks. Through categorizing the stress levels and visualizing the changes in stress levels among participants in the control group and intervention group, we used machine learning algorithms to predict the stress level of non-responding days.
Research Methods: Survey, Quantitative Analysis -
An Agent-Based Modeling Approach for Informing the U.S. Plastic Waste Management Process
Recycling is one of the most significant issues in the waste management system. As the use and demand for plastics increase every year, finding efficient and environment-friendly solutions to handle the plastics in the plastic waste management system gets more challenging. There are economic, environmental, and educational factors affecting plastic waste management. This paper investigates the effects of educational campaigns and system-wide improvement. For this, we used an Agent-Based Modeling and Simulation approach in the NetLogo environment. We provided various scenarios in the current plastics waste life cycle using a real dataset to validate our model, which was from the American Chemistry Council and the National Association for PET Container Resources from 2018. We found that education, technology, and infrastructure changes should be considered holistically to overcome this problem at a system level.
Research Methods: Modeling and Simulation, Qualitative AnalysisHuang, Y., Karabiyik, T., Madamanchi, A., & Magana, A.J. “An Agent-Based Modeling Approach for Informing the U.S. Plastic Waste Management Process,” in International Journal on Advances in Software, 2021, p. 65 to 71.
Our poster won the second-runner at the meeting of Future Work and Learning 2021, West Lafayette, IN