76
Although the medical artificial intelligence (AI) community intensively debates algorithmic fairness and data bias,1–3 these crucial discussions implicitly assume a foundation of existing diagnostic infrastructure. However, a more fundamental challenge is being overlooked: 47% of the world’s population lacks access to even basic diagnostics.4 For these populations, the pivotal issue is not the refinement of diagnosis through AI, but the very existence of diagnostic capacity—a chasm that the current trajectory of AI innovation risks widening.
