Are stroke predictive algorithms accurate regardless of race?

When comparing new stroke-specific algorithms to existing pooled cohort equations for atherosclerotic cardiovascular disease more broadly to estimate the 10-year risk of new-onset stroke, both demonstrated significant discrimination by race (C index 0.76 for White women vs 0.69 for Black women, P<0.001; 0.71–0.72 for White men vs 0.64–0.66 for Black men, P<0.001).
The authors concluded that new machine learning stroke prediction algorithms do not improve discriminative accuracy. Improved modeling techniques are needed to address racial disparities.