🌙 Sleep Health Chatbot

📝 Exploring Personalized Health Support through Data-Driven, Theory-Guided LLMs: A Case Study in Sleep Health (CHI' 25)

|

Contact: 📧 Xingbo Wang (wangxbzb@gmail.com)

This code base demonstrates a LLM-powered sleep health chatbot augmented by a multiagent framework to provide personalized, data-driven, and theory-guided sleep health support. The chatbot integrates wearable device data, contextual information, and established behavior change theories to deliver adaptive recommendations and motivational support.

📁 Project Structure

• ThSleepHealthBot.py - The main file for the chatbot interface and user interaction
• agent_coordinator.py - Coordinates multiple specialized agents (e.g., Recommendation, Health Data Analysis)
• activityRec.py - Contextual multi-armed bandit model for activity recommendation
  • generate_activity - Generate synthetic data for activity recommendations research demonstration
• utils.py - Utility functions for data retrieval and processing
• globalVariable.py - Global variables for the project
• requirements.txt - The dependencies for the project
• README.md - The readme file for the project
• .streamlit - The streamlit secret configuration file: .streamlit/secrets.toml

🚀 Getting Started

📋 Prerequisites

• Python 3.10.13

💻 Installation

1. Create a virtual environment:

python -m venv venv

2. Activate the virtual environment:

source venv/bin/activate

3. Install the dependencies:

pip install -r requirements.txt

⚙️ Environment Setup

Create a .streamlit/secrets.toml file in the .streamlit directory with the following configuration:

• We have provided an example of recommendation model in the parent directory: user_model.pkl

[oura]
oura_token = "input_your_oura_token"

[weatherapi]
weatherapi_key = "input_your_weatherapi_key"

[user]
pid = 1
user_name = "johndoe"
user_model_path = "user_model.pkl"

Required environment variables:

• OPENAI_API_KEY - The API key for the OpenAI API
• oura_token - The token for the Oura API
• weatherapi_key - The API key for the WeatherAPI
• User information:
  • pid - Participant ID
  • user_name - User name
  • user_model_path - Path to the user model file (An example user model is provided)

🏃‍♂️ Running the Chatbot

Run the chatbot using:

streamlit run ThSleepHealthBot.py

📚 Citation

@inproceedings{sleepllm_chi25,
  author = {Wang, Xingbo and Griffith, Janessa and Adler, Daniel A. and Castillo, Joey and Choudhury, Tanzeem and Wang, Fei},
  title = {Exploring Personalized Health Support through Data-Driven, Theory-Guided LLMs: A Case Study in Sleep Health},
  year = {2025},
  isbn = {9798400713941},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  url = {https://doi.org/10.1145/3706598.3713852},
  doi = {10.1145/3706598.3713852},
  articleno = {507},
  numpages = {15},
  series = {CHI '25}
}