Evaluating Conversational Artificial Intelligence for Depression Management
George Mason University
Summary
The goal of this clinical trial is to evaluate how a conversational method of collecting medical history affects patients' perceptions and experiences compared to clinical care as usual. This conversational AI intake system collects medical history information, can be completed by participants at home, and do not disrupt routine clinical care. The primary questions this study aims to answer are: 1\) Does conversational intake affect patients' perceptions of empathy during their clinical interactions? This will be a prospective study that follows a cohort of participants for four (4) months after engaging with the AI intake system. Because each participant serves as his/her own control, both comparators will be administered within-subject, and the order of exposure (AI intake vs. usual care) will be randomized to minimize sequence effects. After completing the AI intake method, participants will rate their experience, particularly in terms of empathy and compare it to their usual interactions with their own clinicians.
Description
Conversational artificial intelligence (AI) systems, such as those based on Large Language Models (LLMs) like ChatGPT, offer innovative ways to engage patients in health-related conversations. Despite these advances, challenges remain regarding patient safety and system reliability. Specific concerns include biased recommendations against certain patient groups, inaccuracies or misleading responses, and mechanical, unempathic interactions, particularly during sensitive moments such as when patients express suicidal thoughts. Testing conversational AI in healthcare settings is complicated due t…
Eligibility
- Age range
- 18–85 years
- Sex
- All
- Healthy volunteers
- No
1. Participant is between 18 to 85 years old. 2. Participant has been, or are likely to be, diagnosed with moderate to severe Major Depressive Disorder without signs of bipolar depression. 3. Participant is not in active suicidal crisis and do not face imminent risk of suicide within the next 3 hours. 4. Participant is not pregnant or seeking to be pregnant. 5. Participant able to communicate in English on the Internet. 6. Participant must reside in the United States. 7. Participant has access to a mental health clinician, or the participant is willing to see study clinicians to help review th…
Interventions
- OtherConversational AI system vs Usual Care
Participants complete medical history intake through an interactive conversational AI designed to support patient-centered, empathetic dialogue. Using large language models (LLM), the system interprets patient input, maintains context, and generates natural-language responses. A dialogue manager prioritizes medically relevant topics to support efficient data collection and reduce off-topic discussion. For safety, trained human monitors oversee conversations in real time and can intervene if risks such as self-harm arise. The AI intake is compared with patients' experiences with their clinicians through monthly follow-up questionnaires over four months. The study evaluates patients' ratings of empathy, communication quality, and engagement, not conversation content. Each participant serves as their own control, with AI intake and usual care compared within-subject and randomized by order of exposure.
Location
- George Mason UniversityFairfax, Virginia