Researchers at George Mason University have found that AI tools used to recommend antidepressants work better when they consider a patient’s race and background.
Their study shows that AI models that ignore race may not give the best advice, especially for African American patients.
This new research was published in the Journal of Health Equity.
It builds on earlier work where the team created an AI system to suggest antidepressants based on medical history. Now, they wanted to see if adding more personal information—like race—would improve the system’s suggestions.
The answer was yes. The team, led by Professor Farrokh Alemi, tested how well AI models worked with and without knowing a patient’s race.
They found that the “race-blind” models—those that didn’t know a patient’s race—often recommended drugs that didn’t work as well for African American patients. On the other hand, models that included race and factors common among African Americans gave much better recommendations.
Vladimir Cardenas, a master’s student involved in the study, said, “The AI models that used race-specific information gave better antidepressant suggestions every time. This shows that just like doctors, AI should consider each patient’s unique background, especially for African Americans with depression.”
Professor Alemi added that AI can be wrong if it doesn’t use all the right information. “If we leave out race and other key factors, the AI might suggest treatments that are less effective for certain groups,” he said.
The researchers stressed that race is not a biological reason for depression or how someone responds to medication.
Instead, it’s linked to other social and environmental factors. For example, many African American patients face challenges like poverty, low access to mental health care, violence, or stigma around getting help. These factors can affect how well treatments work.
The AI system used in this study made suggestions based on medical history and whether the patient had completed their medication or stopped taking it early. Stopping medication was seen as a sign that the treatment wasn’t working.
By adding race-related context, the researchers showed that AI can be made smarter and more helpful. “We want AI to work for everyone,” said Cardenas. “Doctors already try to make personalized choices for their patients. If designed right, AI can support them in doing that.”
The study sends a strong message: for AI in health care to truly help all people, it must consider the full picture, including personal and social factors like race.
If you care about depression, please read studies about how dairy foods may influence depression risk, and B vitamins could help prevent depression and anxiety.
For more information about mental health, please see recent studies that ultra-processed foods may make you feel depressed, and extra-virgin olive oil could reduce depression symptoms.
Copyright © 2026 Knowridge Science Report. All rights reserved.
