Rachael King

Graduate Trainee, George Mason University

1 active project

HAP 823 Course Project (Registered)

Examine the accuracy of an Artificial Intelligent (AI) guide for management of major depressive disorder, including selection of optimal oral antidepressant. This proposal evaluates the impact of this aid on (a) clinicians’ prescription patterns and (b) patients’ depression-free days. R21…

Scientific Questions Being Studied

Examine the accuracy of an Artificial Intelligent (AI) guide for management of major depressive disorder, including selection of optimal oral antidepressant. This proposal evaluates the impact of this aid on (a) clinicians’ prescription patterns and (b) patients’ depression-free days.
R21 Aim 1: Stress-test the conversational AI intake. Generative AI systems have been shown to make false and culturally insensitive statements. Part of this aim is to modify the large-language model to prevent inappropriate responses and to ensure complete medical history intake.
R21 Aim 2: Retrospectively test the accuracy of AI’s advice for African Americans and Hispanics.
R33 Aim 3: Prospectively evaluate the effectiveness of the decision aid.
This study is likely to improve use and effectiveness of antidepressants, including for African Americans and Hispanics, groups that are often excluded in medication effectiveness studies.

Project Purpose(s)

  • Educational

Scientific Approaches

Several new features will be added to the AI system to improve safety. First, the dialogue management will be assisted with a real-time, human monitoring. The dialogue management system sends both the patient’s and the LLM’s deidentified exchanges to a trained monitor. When the trained monitor is not available, the AI system is temporarily closed. If the patient expresses suicidal intent, the human observer can instruct the dialogue manager to transfer the interview to the project’s on-call clinician. In Aim 2 we will evaluate impact of AI Guide using a retrospective analysis of existing data. We will use data available through All of Us Research Hub, coordinated by NIH. A hierarchical analysis of variance is used to see if additional risk factors for minority patients predict response to antidepressants, after controlling for predictions of race-neutral model. We will conduct a cross-over, cluster, practice-based, randomized trial across 2 time periods.

Anticipated Findings

This study is likely to improve use and effectiveness of antidepressants, including for African Americans and Hispanics, groups that are often excluded in medication effectiveness studies.

Demographic Categories of Interest

  • Race / Ethnicity

Data Set Used

Registered Tier

Research Team

Owner:

Collaborators:

  • Vladimir Cardenas - Graduate Trainee, George Mason University
1 - 1 of 1
<
>
Request a Review of this Research Project

You can request that the All of Us Resource Access Board (RAB) review a research purpose description if you have concerns that this research project may stigmatize All of Us participants or violate the Data User Code of Conduct in some other way. To request a review, you must fill in a form, which you can access by selecting ‘request a review’ below.