Weighing the benefits and risks of collecting race and ethnicity data in clinical settings for medical artificial intelligence
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Many countries around the world do not collect race and ethnicity data in clinical settings. Without such identified data, it is difficult to identify biases in the training data or output of a given artificial intelligence (AI) algorithm, and to work towards medical AI tools that do not exclude or further harm marginalised groups. However, the collection of these data also poses specific risks to racially minoritised populations and other marginalised groups. This Viewpoint weighs the risks of collecting race and ethnicity data in clinical settings against the risks of not collecting those data. The collection of more comprehensive identified data (ie, data that include personal attributes such as race, ethnicity, and sex) has the possibility to benefit racially minoritised populations that have historically faced worse health outcomes and health-care access, and inadequate representation in research. However, the collection of extensive demographic data raises important concerns that include the construction of intersectional social categories (ie, race and its shifting meaning in different sociopolitical contexts), the risks of biological reductionism, and the potential for misuse, particularly in situations of historical exclusion, violence, conflict, genocide, and colonialism. Careful navigation of identified data collection is key to building better AI algorithms and to work towards medicine that does not exclude or harm marginalised groups.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it