Research Direction

SEA Lab mainly focuses on two groups of people: (i) general populations, including end-users and patients, to promote health and well-being, and (ii) health experts, including physicians, nurses, and other healthcare providers, to support improved decision-making.

Targeting at these two user groups, the mission of SEA Lab includes three core areas:

1 - Reliable Multimodal Health Insights and Prediction Models

We leverage data from everyday sensors (smartphones, wearables, IoT, social media, to name just a few) and medical records to extract novel behavior insights and develop reliable AI/ML models for health prediction. This supports both general population and health experts.
Mental-LLM: Leveraging Large Language Models for Mental Health Prediction via Online Text Data
2024. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies.
Cite DOI URL
Mental-LLM: Leveraging Large Language Models for Mental Health Prediction via Online Text Data
Health-LLM: Large Language Models for Health Prediction via Wearable Sensor Data
2024. Proceedings of Machine Learning Research.
Cite URL
Health-LLM: Large Language Models for Health Prediction via Wearable Sensor Data
GLOBEM: Cross-Dataset Generalization of Longitudinal Human Behavior Modeling
2023. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies.
Cite DOI URL
GLOBEM: Cross-Dataset Generalization of Longitudinal Human Behavior Modeling
See all

2 - Intelligent Interaction and Intervention Techniques

Building on predictive models, we further design and develop human-AI interaction and behavior change intervention techniques that are aimed at promoting health and well-being. This area mainly focuses on general populations.
Time2Stop: Adaptive and Explainable Human-AI Loop for Smartphone Overuse Intervention
2024. Proceedings of the CHI Conference on Human Factors in Computing Systems.
Cite DOI URL
Time2Stop: Adaptive and Explainable Human-AI Loop for Smartphone Overuse Intervention
MindShift: Leveraging Large Language Models for Mental-States-Based Problematic Smartphone Use Intervention
2024. Proceedings of the CHI Conference on Human Factors in Computing Systems.
Cite DOI URL
MindShift: Leveraging Large Language Models for Mental-States-Based Problematic Smartphone Use Intervention
InteractOut: Leveraging Interaction Proxies as Input Manipulation Strategies for Reducing Smartphone Overuse
2024. Proceedings of the CHI Conference on Human Factors in Computing Systems.
Cite DOI URL
InteractOut: Leveraging Interaction Proxies as Input Manipulation Strategies for Reducing Smartphone Overuse
See all

3 - AI-Powered Clinical Decision Support

We integrate AI models into clinical decision support systems, enabling health experts to interpret actionable insights, collaborate with AI, and make more informed decisions. This area is experts-focused.
Talk2Care: An LLM-based Voice Assistant for Communication between Healthcare Providers and Older Adults
2024. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies.
Cite DOI URL
Talk2Care: An LLM-based Voice Assistant for Communication between Healthcare Providers and Older Adults
From Classification to Clinical Insights: Towards Analyzing and Reasoning About Mobile and Behavioral Health Data With Large Language Models
2024. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies.
Cite DOI URL
From Classification to Clinical Insights: Towards Analyzing and Reasoning About Mobile and Behavioral Health Data With Large Language Models
Rethinking Human-AI Collaboration in Complex Medical Decision Making: A Case Study in Sepsis Diagnosis
2024. Proceedings of the CHI Conference on Human Factors in Computing Systems.
Cite DOI URL
Rethinking Human-AI Collaboration in Complex Medical Decision Making: A Case Study in Sepsis Diagnosis