CRII: HCC: Advancing Trauma-Informed Care in AI-driven Mental Health Chatbots

Parkview Hospital, Inc., Fort Wayne IN

Investigators

Abstract

The use of artificial intelligence (AI)-based conversational agents, or "chatbots", is a promising way to make mental health support more readily available to people who need it. However, the AI models designed to generate conversation with users can make mistakes, and thus need to be designed to minimize the risk of giving incorrect advice, or using language that might harm vulnerable people with mental health needs. This project's goal is to use trauma-informed care (TIC) principles in the design of mental health chatbots to reduce those risks. TIC is an important approach in mental health support that prioritizes patient safety, trust, peer support, collaboration, empowerment, and cultural/historical sensitivity, holding providers and healthcare organizations accountable during patient care. This project will use TIC principles, along with guidance from both mental health providers and chatbot users, to assess existing chatbots for mental health support as well as how they fit into mental health practice. The team will develop guidelines that embody the TIC framework in chatbots to prioritize user safety, empowerment, and wellbeing, as well as giving more general insight into the application of TIC principles in healthcare technologies. By establishing accountable design practices for AI-driven mental health chatbots, this project seeks to improve care quality, user experiences, and scientific progress in the field. While the application of AI-based chatbots has significantly increased in health domain, there is no specific regulation and accountability for these platforms to ensure TIC principles are incorporated. The project aims to assess human-chatbot interactions through the lens of trauma-informed care, employing a comprehensive mixed-methods approach that triangulates data from user surveys, semi-structured interviews, and interactive group sessions with providers. This project has three primary objectives. First, the team will systematically explore user perceptions of mental health chatbots through surveys and semi-structured interviews, shedding light on user engagement, expectations, satisfactions, and concerns while seeking support. Second, the team will engage multidisciplinary mental health professionals in interactive group sessions with existing mental health chatbots, characterizing interactions from a trauma-informed care perspective and fostering shared recommendations for chatbot design. Third, the team will develop guidelines for human-chatbot interaction strategies, rooted in human-centered design methods and TIC principles, with the aim of establishing a foundation for accountable and safe AI and chatbot usage in the domain of mental health. By disseminating findings within scientific communities, the research team aims to raise awareness about the importance of translating insights between computing and clinical research. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.