📋

Pre-Registered Research

This study's methodology and analysis plan were specified before data collection began, preventing post-hoc adjustment of methods to achieve desired results.

Protocol version 1.0 — January 2026

Data Downloads

All datasets in JSON format for maximum accessibility.

👥

Name Pairs Dataset

54 name pairs across 6 demographic contrast categories with length matching and category labels.

JSON 11 KB 54 pairs
Download JSON
🏥

Symptom Profiles

20 standardized symptom profiles covering cardiovascular, neurological, abdominal, respiratory, pain, mental health, and fatigue categories.

JSON 10 KB 20 profiles
Download JSON
💢

Pain Vignettes

5 clinical reasoning vignettes for pain management scenarios with demographic variants.

JSON 6 KB 5 vignettes
Download JSON
❤️

Cardiac Vignettes

5 clinical reasoning vignettes for cardiac scenarios including STEMI, atypical chest pain, and heart failure.

JSON 6 KB 5 vignettes
Download JSON
🧠

Psychiatric Vignettes

5 clinical reasoning vignettes for psychiatric scenarios including psychosis, depression, and anxiety.

JSON 6 KB 5 vignettes
Download JSON
🚨

Emergency Vignettes

5 clinical reasoning vignettes for emergency scenarios including acute abdomen and mental health crisis.

JSON 6 KB 5 vignettes
Download JSON

Key Literature

Foundational research our methodology builds upon.

Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019)

Dissecting racial bias in an algorithm used to manage the health of populations

Science, 366(6464), 447-453

Relevance: Demonstrated racial bias in a healthcare algorithm affecting millions of patients. Foundational methodology for algorithmic fairness research.
DOI: 10.1126/science.aax2342 →

Hoffman, K. M., Trawalter, S., Axt, J. R., & Oliver, M. N. (2016)

Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites

PNAS, 113(16), 4296-4301

Relevance: Documented systematic racial bias in pain assessment. Our pain management findings mirror these human biases.
DOI: 10.1073/pnas.1516047113 →

Schulman, K. A., et al. (1999)

The effect of race and sex on physicians' recommendations for cardiac catheterization

NEJM, 340(8), 618-626

Relevance: Classic study showing disparities in cardiac referrals. Our cardiac vignette findings show similar patterns in AI systems.
DOI: 10.1056/NEJM199902253400806 →

Bertrand, M., & Mullainathan, S. (2004)

Are Emily and Greg More Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination

American Economic Review, 94(4), 991-1013

Relevance: Established the matched-pair testing methodology using names as demographic signals that our research extends to healthcare AI.
DOI: 10.1257/0002828042002561 →

Strakowski, S. M., et al. (2003)

The impact of race on diagnosis and disposition from a psychiatric emergency service

Journal of Clinical Psychiatry, 64(4), 395-401

Relevance: Documented racial disparities in psychiatric diagnosis (schizophrenia vs. bipolar). Our psychiatric findings parallel this pattern.
View Study →

Citation

How to Cite This Research

Healthcare AI Fairness Observatory (2026). Name-Based Discrimination in Consumer Medical AI Symptom Checkers: A Matched-Pair Experimental Study. https://aifairnesslab.org
Additional Formats

License

💻 Code

MIT License

Free to use, modify, and distribute. Include original license and copyright notice.

📊 Data

CC-BY-4.0

Free to share and adapt. Attribution required.

📝 Documentation

CC-BY-4.0

Free to share and adapt. Attribution required.

Questions?

If you have questions about the research, data, or resources, please reach out.