DINT #70 - Does AI Make Healthcare Racism More Prevalent?

Plus: India does an about-face on its computer hardware import ban.

News Briefs

Microsoft Ends Support for the Standalone Cortana App, Making Way for AI in M365 (Thurrott / Laurent Giret)

Cyberattack Hits Hospitals in Pennsylvania and New York, Disrupting Treatment (The Associated Press / Pat Eaton-Robb)

Apple Hits the 1B Subscription Mark Across All of Its Digital Platforms (Axios / Sara Fischer)

Ponzi Schemers Target the Unemployed in Zimbabwe (rest of world / Kimberly Mutandiro)

India Presses Pause on Its Laptop Import Ban (TechCrunch / Manish Singh)

---

---

Friday Feature

We’re deluding ourselves if we believe new technology will somehow clean the slate and absolve us of the racism inherent in every facet of our society, especially healthcare.

Today Microsoft announced a partnership with Duke University to provide AI infrastructure for the university’s healthcare arm, Duke Health. The partnership will provide an exchange between the organizations with Duke Health providing Microsoft with healthcare use cases and baseline data for analysis. Microsoft will give Duke Healthcare the chance to leverage AI, LLMs, and emerging technology best practices to gather key information to make the partnership fruitful.

AI hasn’t been friendly to people of color when seeking unbiased treatment. Here’s a quick rundown of what Black and Brown people face:

• Higher maternal mortality rate

• Less comprehensive care for serious medical issues

• Skewed systems that misdiagnose or put Black and Brown patients at a higher risk of death

• Misdiagnosis due to medical evaluation tools that aren’t calibrated to detect healthcare issues for people with dark(er) skin

All four issues existed before the advent of AI. The technology now makes it easier to track and, one hopes, eliminate from the healthcare structure.

When looking at the Duke/Microsoft partnership, the words ‘ethically harness’ are used to describe the approach of the partnership. A closer look at the partnership reveals a baseline awareness of medical bias fueled by algorithms that reflect the systemic racism in our society.

"Adoption of AI in [healthcare] must avoid previous pitfalls related to clinical bias (racial bias in transplant recommendations, gender bias in screening for liver disease, skin color bias in diagnosis, etc.); ensure that ethics are a cornerstone of care delivery; vet model inputs and reliability and thoroughly test recommendations," said UNC Health’s Chief Analytics Officer Rachini Ahmadi-Moosavi to Healthcare IT News.

Can Big Tech companies like Microsoft be trusted to provide ethical generative AI resources? The firm touts its Frontier Model Forum with OpenAI, Google, and Anthropic to regulate AI. AI ethicist Timnit Gebru says this about Big Tech self-monitoring ethics:

"… unless there is external pressure to do something different, companies are not just going to self-regulate. We need regulation and we need something better than just a profit motive,” Gebru told The Guardian last May.

Gebru penned a landmark paper on algorithmic bias during her days at Google. The key takeaway for many was the ‘black box’ of AI in which systems are trained on so much data that it’s difficult to hold someone or something accountable for biased or erroneous information.

One algorithm threatens the health outcomes of millions of people of color.

Racial Bias in Health Algorithms

Authors of a landmark study on algorithmic bias in healthcare revealed troubling trends that point to how decisions outside of tech, around healthcare costs, provide a pathway to racial bias. Here’s an excerpt from the study, “Dissecting racial bias in an algorithm used to manage the health of populations” by Ziad Obermeyer, Brian Powers, Christine Vogeli, and Sendhil Mullainathan: