Racial bias is built into analytics software designed to target services to the neediest patients, according to a STAT investigation that uncovered the troubling pattern in multiple algorithms used by US hospitals.
- For example, such algorithms use medical spending data as a proxy for health need—disregarding the reality that people of color may have lower care costs because they get fewer tests and check-ups for conditions like heart failure and diabetes.
The impact: “unintentional redlining,” according to a roundup of the findings.
- Just one example: A white patient was given priority access to services after being ranked 4X higher than a Black patient with similar ailments.
Most of these products fail to acknowledge potential racial bias—some even claim to be designed to avoid it, STAT reports in another article.