Back

Reporting of Fairness Metrics in Clinical Risk Prediction Models: A Call for Change

Rountree, L. I.; Lin, Y.-T.; Liu, C.; Salvatore, M.; Admon, A.; Nallamothu, B. K.; Singh, K.; Basu, A.; Mukherjee, B.

2024-03-18 public and global health
10.1101/2024.03.16.24304390 medRxiv
Show abstract

Clinical risk prediction models integrated in digitized healthcare systems hold promise for personalized primary prevention and care. Fairness metrics are important tools for evaluating potential disparities across sensitive features in the field of prediction modeling. In this paper, we seek to assess the uptake of fairness metrics in clinical risk prediction modeling by conducting a scoping literature review of recent high impact publications in the areas of cardiovascular disease and COVID-19. Our review shows that fairness metrics have rarely been used in clinical risk prediction modeling despite their ability to identify inequality and flag potential discrimination. We also find that the data used in clinical risk prediction models remain largely demographically homogeneous, demonstrating an urgent need for collecting and using data from diverse populations. To address these issues, we suggest specific strategies for increasing the use of fairness metrics while developing clinical risk prediction models.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.