Selected Projects

Click on the [click] to see more details.

personalizing Just-In-Time Adaptive Intervention

The personalizing Just-In-Time Adaptive Intervention (pJITAI) Toolbox, is a no-code web-based platform to allow mobile health researchers to create, configure, and deploy Reinforcement Learning (RL) algorithms for mHealth interventions. [click]

Human Support Robot & Pick-and-Place

Fine-tuned the YOLO algorithm to enable real-time object detection for the Human Support Robot (HSR), and enhanced the accuracy of the HSR’s pick-and-place tasks to 90% by implementing a dual-camera system (head and gripper cameras) compared to a single-camera setup. [click]

Personalized Chatbot Design for mHealth

We explore the possibility of customizing Digital Behavior Change Interventions (DBCIs) based on DISC personality profiles, with the goal of aligning the intervention content to the more stable aspects of individual motivational differences.

EDBooks

We suggest methods to integrate large language models (LLMs) with traditional learning materials, such as raw HTMLs, to provide readers the advantages of LLMs—especially the ability to ask personally relevant questions and get tailored responses—while retaining the benefits of conventional raw HTML resources.

Bandit-Supporting Care Planning

To enhance care quality through personalized care planning and delivery, even with a limited workforce, we propose a new AI-assisted care planning model. This model leverages bandit algorithms to optimize clinical decisions by continuously adapting to feedback from previous decisions in real time.

AICAS 2023