Reducing Racial Disparities in Stroke Thrombolysis Using Fairness-Aware Machine Learning. | ScholarDeck

Paper

Reducing Racial Disparities in Stroke Thrombolysis Using Fairness-Aware Machine Learning.

New research · AI in Medicine

Journal of racial and ethnic health disparities · 1d

AI / informaticsJournal of racial and ethnic health disparities· Aug 2026

Reducing Racial Disparities in Stroke Thrombolysis Using Fairness-Aware Machine Learning.

Jean-Luc K Kabangu, Verónica Ramíerz Lopera, Teresia M Perkins … Sonia V Eden

Read paper

AI Evaluation & GovernanceAI / informaticsQ1IF 3.6

Fairness-aware AI cut racial disparities in stroke clot-busting drug use by 64.8%

Reducing Racial Disparities in Stroke Thrombolysis Using Fairness-Aware Machine Learning.

Jean-Luc K Kabangu … Sonia V Eden

Journal of racial and ethnic health disparities · Aug 2026

Background

Racial and ethnic disparities in the administration of intravenous thrombolysis (intravenous thrombolysis) for acute ischemic stroke (acute ischemic stroke) remain persistent, raising concerns that bias-both systemic and algorithmic-may influence treatment decisions.

Methods

Using differential item functioning (differential item functioning) analysis on a national inpatient cohort of 983,785 patients, we found that minority patients required higher NIH Stroke Scale scores than White patients to receive intravenous thrombolysis, indicating potential bias in score interpretation.

n = 983,785 patients

Results

64.8%

AI-driven bias correction narrowed the racial gap in clot-busting drug treatment

n = 983,785 patients

“

Conclusion

Our findings illustrate how standardized tools can contribute to inequity and demonstrate the potential of ethical AI to mitigate disparities in real-world stroke care.

Read paper