Algorithmic discrimination and health equity

Excerpt

Algorithms hold great promise for healthcare. If implemented properly, these technologies can boost outcomes and improve the quality of care, perhaps even reducing disparities in the process. Yet, to varying degrees, algorithms can also discriminate. This chapter considers four potential sources of algorithmic discrimination and explores how they might impact health equity: (1) discrimination from design; (2) discrimination from biased inputs; (3) discrimination from data deficits; and (4) ‘discrimination’ from actuarially sound disparate impacts. Unfortunately, existing antidiscrimination laws in the United States are ill-equipped to address these concerns. We therefore conclude with a series of policy suggestions to minimise discrimination by algorithms, ensuring that they offer as much benefit to as many people as possible.