THURS-093 - Lessons Learned from Building an Ethical and Equitable Ai-based Chatbot for Health Coaching

Brief Abstract Summary: Learn about steps used to adapt an evidence-based health program to an AI-based coaching chatbot, including literature review, human-centered design workshops, framework and strategy development, crafting app and technology flowcharts, prototype development, and iterative user testing and refinement. Ascertain how to incorporate design features for building an ethical chatbot that prioritizes transparency, integrity, and trust to facilitate confidence and safety, a supportive connection between the user and chatbot, and successful behavior change. Gain knowledge about how to develop an equitable, user-centered AI-based platform for behavior change for multicultural populations. Hear about the interdisciplinary collaboration between the public health education team and computer scientists, discovering best practices for communication, project management, and product development for cross-disciplinary experts and teams.

Detailed abstract description: With the advent of AI, health professionals and software developers are rushing to incorporate AI into many aspects of human-based interactions. Incorporating AI recklessly has led to numerous challenges with serious consequences for peoples’ health and well-being. Media coverage has highlighted injuries to individuals from chatbot interactions, but conscientious design and inclusive construction can mitigate potential harms.

In this presentation, we will share our approach to ethically, equitably, and conscientiously building an AI-based chatbot that provides digital health coaching to individuals and helps them make progress towards meeting SMART goals for health behavior change. The chatbot was developed from an evidence-based health and wellness coaching program based in motivational interviewing, social cognitive theory, and the transtheoretical model of change. The chatbot came about, in part, from suggestions and interest following our coaching team’s highly attended skill-building workshop on digital health and wellness coaching, that was presented at the 2025 SOPHE annual conference. Our team is excited to share this new technology-based adaptation of the coaching program at next year’s SOPHE conference.

The adaptation and build process for the chatbot were informed by broader digital health models such as the e-health literacy framework, AI-specific theoretical frameworks such as the AI chatbot behavior change model, and a review of the literature on technology-based coaching interventions. Design features promoting ethical use and equitable implementation were incorporated into the chatbot to build trust, transparency, and integrity; relational capacity such as empathy and caring tone and benevolence in messaging; user personalization and tailoring; and persuasive ability to motivate positive health and well-being outcomes. Attention to equity in the chatbot development was driven by the multicultural team, training on 50+ transcripts from minority participants in the original health coaching program, and the human-centered design framework that provides the ability for each individual user to choose their unique goals and share their lifestyle and behavior preferences to inform tailored support and conversations with the chatbot.

In this presentation, we plan to present the following:
Methodology used to design, prototype, build, and deploy the app in the real world;
Specific design considerations made when building an AI-based health coaching chatbot;
Challenges faced when building the first prototype of the app, and results from prototype testing with users; and
Benefits of working with a diverse team of experts from public health, communication, and computer science.