Including Vulnerable Populations in the Assessment of Data From Vulnerable Populations

Abstract

Data science has made great strides in harnessing the power of big data to improve human life across a broad spectrum of disciplines. Unfortunately this informational richesse is not equitably spread across human populations. Vulnerable populations remain both under-studied and under-consulted on the use of data derived from their communities. This lack of inclusion of vulnerable populations as data collectors, data analyzers and data beneficiaries significantly restrains the utility of big data applications that contribute to human well-ness. Here we present three case studies: (1) Describing a novel genomic dataset being developed with clinical and ethnographic insights in African Americans, (2) Demonstrating how a tutorial that enables data scientists from vulnerable populations to better understand criminal justice bias using the COMPAS dataset, and (3) investigating how Indigenous genomic diversity contributes to future biomedical interventions. These cases represent some of the outstanding challenges that big data science presents when addressing vulnerable populations as well as the innovative solutions that expanding science participation brings.

Keywords: African Americans; COMPAS; Samoa; algorithmic fairness; criminal justice; genomics; inclusion.