Seol mar théacs é seo: A translational perspective towards clinical AI fairness