My research lies at the intersection of human-computer interaction (HCI), and social computing. In specific, my study focuses on how we can design technologies to encourage individuals to engage proactively and become a pro-social actor in online communities. Expressing and feeling empathy in conversations can be a powerful motivating factor for people to engage in social activities. I believe that effective representation of context-relevant information can facilitate effective communication among people with the help of mediated technologies. Currently, I am interested in investigating effective communication strategies in personal health informatics and collaborative work in healthcare and classroom settings respectively. From theory to practice, I have committed to conceptualize user behavior model and suggest dyadic and triadic interaction mechanisms deeply integrated into human-artifact and environment.

Current Project
NIH R21: Real-Time Prediction of Marijuana Use - Effects of Use on Cognition in the Natural Environment

This NIH R21 project aims to address limitations of existing research by (1) developing an algorithm to predict MJ use in real-time using smartphone data, and (2) examining effects of marijuana use on cognition using smartphone-based cognitive testing in the natural environment. Development of an algorithm to predict marijuana use in real-time would, among its many healthcare and research applications, facilitate systematic assessment of marijuana effects on cognitive functioning through more efficient scheduling of smartphone cognitive testing in the natural environment.

Collaborators: Tammy Chung (UPitt, PI), Brian Suffoletto, Anind Dey
Role: CO-PI

NIAAA CTSIA Text-Message Intervention to Reduce Alcohol Use among Young Adult ED Patients

Young adults have the highest rate of hazardous alcohol consumption and suffer disproportionately from alcohol related injuries. NIAAA funded an R01 for us to figure out (1) what Text Message components are needed for change and (2) mechanisms through which they change. Starting to understand mechanisms through which individuals change drinking behavior.

Collaborators: Brian Suffoletto (UPitt, PI), Tammy ChungAnind Dey
Role: Co-I
Publication: Addictive Behavior, 2018 [pdf]

National Center for Advancing Translational Sciences (NCATS): Sensor-Based Markers of Drinking and Impairment

This grant aims to develop a model for real-time prediction of alcohol use in young adult hazardous drinkers using phone sensor-based markers.

Collaborators: Brian Suffoletto (UPitt, PI), Tammy ChungAnind Dey
Role: Co-I
Proceedings of the ACM on  Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2017 [pdf]

Biobehavioral Predictors of Recovery after Cancer Surgery

We conducted the research to see if behavioral patterns can be tracked using smartphone and wearable sensors predict clinical outcomes such as readmission after cancer surgery. We investigated to see if we can use mobile technology to reduce sedentary behavior before and after cancer surgery, and improve postoperative health.

Collaborators: Carissa Low (UPitt, PI)Anind Dey

Role: Co-I
ACM International Joint Conference on Pervasive and Ubiquitous Computing , 2016 [pdf]; JMIR, 2017 
Past Projects
Home Health Monitoring System for Preventative Care

We investigated how elders can self-monitor their vital signs using consumer health products and whether they could comply with their treatment plan successfully. The goal of this study is to develop a system to provide guidance to recently discharged CHF patients and prevent their readmission to the hospital.

Collaborators: Jodi Forlizzi, John Zimmerman, Steven Dow

Conflict Management in Group Collaborative Setting

Kiva is a collaboration tool that facilitates group communication and helps organize project work. The Kiva is based on the idea that groups communicate through posts; that is, the Kiva feels like a chat session in which group members send and receive messages to one another. We looked at how group members solve problems within and outside the work group.

Collaborators: Dan Siewiorek, Asim Smailagic and Jodi Forlizzi

Context Information in Location Based Social Network Services

For my dissertation work, I studied how and why users of social networking services (SNS) provide social support to anonymous SNS users. A typical posting on a location-based SNS app comprises four different types of context information: physical activity (e.g. walking), location (e.g. wall st.), time (e.g. 2:00 pm), physical environment (e.g. quiet), and emotion (e.g. happy). Any combination of these information types can be expressed in two forms: subjective (e.g. I am moving slowly. My surroundings are quiet) and objective (e.g. Movement 1 km/h. Surrounding noise level 10 dB). I looked at how people respond to an anonymous user's posting in an LBSNS mobile app that were controlled for the type of context information and form expression.

Advisor: Jinwoo Kim
Role: Co-I
Publication:  International Journal of Human-Computer Studies (IJHCS) 2013, [pdf]

The effects of egocentric and allocentric representations on presence and perceived realism

This study aims to construct a theoretical model that explains the perceived effects of stereoscopic 3D features on sense of presence, and to verify the validity of the model in the 3D computer game domain. The study focuses on spatial representation and perceived realism as important mediating factors between the perceived system features and sense of presence. According to the Dual Mode Model (DMM), two types of spatial representation are crucial for perceived realism and presence: egocentric representation and allocentric representation. 

Advisor: Jinwoo Kim
Role: Co-I
Publication: Interacting with Computers 2012, [pdf]