Mitigating Bias in Healthcare Algorithm Development

News
Article

A panel designs a conceptual framework in an effort to limit partiality.

Image Credit: Adobe Stock Images/Infografx.com

Image Credit: Adobe Stock Images/Infografx.com

Healthcare algorithms are designed to use mathematics not only to help boost decision-making abilities but improve health outcomes in the process. Despite these well-intentioned efforts, history has shown that algorithmic bias has plagued minorities in banking, education, and housing sectors, and the same holds true for healthcare.1

In fact, it is of such importance that the Biden Administration issued an Executive Order noting that “agencies shall consider opportunities to prevent and remedy discrimination, including by protecting the public from algorithmic discrimination.”2

To address this issue further, the Agency for Healthcare Research and Quality (AHRQ) and the National Institute on Minority Health and Health Disparities (NIMHD) assembled a nine-person panel of experts with varying backgrounds and knowledge that would offer “core guiding principles” for the design and use of clinical algorithms in the healthcare sector; the group would examine evidence, communicate with stakeholders, and collect feedback from the community.

The principles would feature data-driven algorithms that use artificial intelligence (AI) and machine learning. It’s also important to note that these aforementioned principles would even pertain to any rules-based approaches that are derived from data.

A study published in JAMA Network Open3 describes the process for mitigating this algorithmic bias by attempting to control it at each lifecycle phase. The lifecycle of an algorithm consists of five phases that usually occur chronologically:4

  • Defining and forming the problem
  • Data selection, assessment, and management
  • Algorithm development, training, and validation
  • Integrating these algorithms in desired settings
  • Monitoring the algorithm and making adjustments

The conceptual framework determined that these phases should be defined by the following principles, or guidelines that can serve as reminders throughout the process:

  1. Promoting health and healthcare equity throughout all phases of the healthcare algorithm lifecycle
  2. These algorithms (and their use) should be transparent and able to be explained
  3. Patients and communities need to be engaged during the entire lifecycle in order to establish trustworthiness
  4. Clearly determine any issues surrounding bias
  5. Hold these algorithms accountable for providing equity and fairness in determined outcomes

The prominence of AI models—including ChatGPT—have shined a light on both the positives and negatives of algorithms. As a result, the article suggests that various parties collaborate in order to design solutions that combat these obstacles, whether that be in the shape of incentives, processes, regulations, standards, systems, and policies.5

“Dedicated resources and the support of leaders and the public are critical for successful reform,” the authors wrote. “It is our obligation to avoid repeating errors that tainted use of algorithms in other fields.”

References

1. O’Neil, C. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown; 2016.

2. The White House. Executive Order on further advancing racial equity and support for underserved communities through the federal government. Updated February 16, 2023. Accessed August 31, 2023. https://www.whitehouse.gov/briefing-room/presidential-actions/2023/02/16/executive-order-on-further-advancing-racial-equity-and-support-for-underserved-communities-through-the-federal-government/

3. Chin, MH, Afsar-Manesh, N, Bierman AS, et al. Guiding Principles to Address the Impact of Algorithm Bias on Racial and Ethnic Disparities in Health and Health Care. JAMA Netw Open. 2023;6(12):e2345050. doi:10.1001/jamanetworkopen.2023.45050

4. Ng, MY, Kapur, S, Blizinsky, KD, Hernandez-Boussard, T. The AI life cycle: a holistic approach to creating ethical AI for health decisions. Nat Med. 2022;28(11):2247-2249. doi:10.1038/s41591-022-01993-y

5. Cary, MP Jr, Zink, A, Wei S, et al. Mitigating racial and ethnic bias and advancing health equity in clinical algorithms: a scoping review. Health Aff (Millwood). 2023;42(10):1359-1368. doi:10.1377/hlthaff.2023.00553

Recent Videos
© 2024 MJH Life Sciences

All rights reserved.