Members
- Behnam Rahdari – Postdoctoral Scholar with expertise in personalization, and recommender systems. Now working on clinician-AI interaction using LLMs.
- Rashon Poole – Student Research Assistant with expertise in full stack development and prompt engineering for LLMs. Currently working on diabetes data decision-support.
- Elahe Bashiri – Research Associate with expertise in qualitative research. Currently working on examining clinician-data scientist collaboration for the development of AI/ML solutions in healthcare.
- Hee Jung Choi - Student Research Assistant with expertise in machine learning and prompt engineering for LLMs. Currently working on diabetes data decision-support.
Alums
- Till Scholich - Senior PM at FICUS Health. Previously worked on comparing LLMs to licensed therapists and on designing a diabetes decision-support dashboard for clinicians. Led two publications.
Projects
Patient-Generated Data
Data generated by patients outside clinical settings carries immense potential in informing decisions. However, the current ecosystem of tools does not help realize the potential of this data. I study the use of this data and design data interfaces to support analysis, interpretation, and decision-making in data-driven care of chronic health conditions. I draw from theories of sensemaking to characterize practices of patients and clinicians and translate those practices to interface features that enable productive engagement with data. We are working on evaluating GPT-5’s ability to interpret data from continuous glucose monitors and insulin pumps with the aim to create multimodal summaries of data for review in the clinic.
AI in Healthcare
Despite millions of dollars being spent in developing artificial intelligence capabilities, evidence shows that AI algorithms and systems that are successful in the lab often fail when deployed in the field - hospitals, and clinics. AI in its current state has been resulting in suboptimal experiences and results, which can be attributed to the lack of focus on the different types of users of AI, the contexts in which AI is embedded, and suboptimal interdisciplinary collaboration through which models are conceptualized and developed. We are studying human interaction with artificial intelligence capabilities in healthcare technology, such as clinical decision-support tools for diagnostics, and patient-facing automated therapy systems to improve how healthcare AI systems are conceptualized, and designed. We are also studying how clinicians and data scientists work together to develop computational solutions.
Mobile Health(mHealth)
In the past, I have worked on several projects related to mHealth. I also recently completed the mHealth training institute in LA where I learned methods for translating and evaluationg mobile health interventions in the field. My past work in this area is related to context-awareness and ecological momentary assessment.
Context and Context-Awareness in Health
To support patients with chronic health conditions, it is important to understand the contexts in which self-care activities are performed and overall, the role of contextual factors in affecting health-related behaviors. I study the lived experiences of people to understand how context can be better utilized in interventions, such as just-in-time interventions, for assisting users in health-related activities. Specifically, I have studied the influence of context on the management of type 1 diabetes and on planning for physical activity. I am also interested in studying practices of designers who create context-based applications using health data.
Self-Reporting and EMA
Many researchers rely on self-reported inputs from their participants to get data. However, self-reporting remains a burdensome activity despite improvements in technology. I have collaborated with several colleagues to explore the use of situated self-reporting devices as compared to mobile phones for self-reporting. We have observed that mobile phones complement the use of stand-alone devices for self-reporting. I am interested in further assessing multi-device set-ups to improve self-reporting technology and experience.