To address the incorporation of race in clinical algorithms and evaluate patient preferences at scale as a key consideration in algorithm design.
Approach:
Concerns with Race Incorporation: Acknowledges methodological problems with incorporating race into clinical algorithms, including clinicians’ misidentification and geographic distribution shifts over time.
Key Findings:
Different race-removal approaches can have varying effects on health disparities, potentially improving or worsening them.
Race often serves as a proxy for unmeasured factors that algorithms cannot currently incorporate, necessitating systemic measures.
Patient attitudes towards race use and rigorous comparisons of algorithms are essential for advancing race-neutral clinical algorithms.
Interpretation:
The authors advocate for a careful and methodologically rigorous approach to the incorporation of race in clinical algorithms, emphasizing the need for systemic changes to address health disparities and improve measurement.
Limitations:
The study does not provide specific methodologies for assessing patient preferences at scale.
It does not detail the systemic measures required to address structural disparities, which are necessary for responsible race removal.
Conclusion:
Advancing toward race-neutral clinical algorithms requires careful assessment, attention to patient attitudes, and systemic efforts to reduce disparities.
